The Absolute Sensitivity of Radio Receivers - North - RCA-Review-1942-Jan

RCA REVIEW
A Quarterly Journal of Radio Progress

Published in July, October, January and April of Each Year by
RCA INSTITUTES TECHNICAL PRESS
A Department of RCA Institutes, Inc.
75 Varick Street, New York, N. Y.
VOLUME VI January, 1942 NUMBER 3
CONTENTS
PAGE
NBC Studios 6A and 6B 259
G. M. NIxoN
Generation and Detection of Frequency -Modulated Waves 269

S. W. SEELEY, C. N. KIMBALL, AND A. A. BARCO
A New Chemical Method of Reducing the Reflectance of Glass 287

F. H. NICOLL
An Analysis of the Signal -to -Noise Ratio of Ultra-High- Frequency
Receivers 302
E. W. HEROLD
The Absolute Sensitivity of Radio Receivers 332
1.1 D. O. NORTH
An Omnidirectional Radio -Range System,

Apparatus
Part II- Experimental
344
D. G. C. LUCK
Measurement of the Slope and Duration of Television Synchronizing

Impulses . 370
R. A. MONFORT AND F. J. SOMERS
Our Contributors 390
Technical Articles by RCA Engineers 392
SUBSCRIPTION :
United States, Canada and Postal Union: One Year $2.00, Two Years $8.00, Three Years $4.00
Other Foreign Countries: One Year $2.85, Two Years $3.70, Three Years $5.05
Single Copies: 75 each
Copyright, 1942, by RCA Institutes, Inc.
Entered as second -class matter July 17, 1936, at the Post Office at New York, New York,
under the Act of March 8, 1879.
Printed in U.S.A.
BOARD OF EDITORS
('Aairn(an
l'H+Mtc, J. I'NIt
Prtatdrrt(. R('A Institutes. loc.
1t.trH It. BtL CHKt r.r, W. IfuKN

Rest-arch Uirectur, .iYS(St(2nI Vice President and
RCA Laboratories Uu<crw uf Research and IeI'rlupltnsnt
\111ntil BrinLtlfUYltny ('uWpuny
UK. H. H. Bk t kJlAt.k
I'r(t PnYtdcut ut ( huryt N'ILl.N1N HI MT
vl s. .st, A wt(! Urt'tluputt str, .iaYlaf/lllt l;ruc ru1 l'ulutltl

R.C..d. Cumin uaic'utions, lsic. Radio ('ury(uution of A ruertcu
ItuukMT S. BUMNr UK. L HAMI.ki B. Jut test 1
Enyist, rr tn Chu/ Vt., Chief Engineer,

r u,utrn rau/ EnyLntrrr uty .'cctlun. RCA Laboratories
l;t .1 ILI n Wu( f (II Lrtg ('wnpuay.
I:u(ltutrunt !Pittston C. W. LAlIIIkY
l' ace Presidt at and (..'hat! Enyuu rr
IMINu F. BMNk.e untntustttUhuns. ill( .
('hiel Enyirteer,
Iiudwrrturutr ('uryurutiun of Antrrtta I''KA N K E. MU' l.b. N
l'ttr Prrsdent and Genet u/ l/antlgrr
Da. AtrRED N. Gut DIM llH Ni, f (044(21 Bruadcustirty ('uUtlut fl
Consulting Engineer,
Iiu,Gu ('ur'purutturt of Anteru'u E. W. ItITiT:t
G. GKwKa
l't(r President In Charge of
HAMMY
Manufacturing and Pt uducfiun
l:rurrul Patent Att urne y, Enyiurrring
Rudtu ('urpurattVn of Awes Ica RCA Manufacturing Company
O. B. HNwN CHKl2, H. Tai

race President in Charge of Enyiltrrrtny R.C.A. ('un /tnunicuttuns, lise.
Nutiurukl Bruudcust iny ('ulst)nkrny
AMTHS M R. VAN DICK
HutluN H. !HAI H
.innl111r5
Uirectur of Adrrrtisuty Industry Service Section,
UM! POMtetly
Radio Corporation of America R('A Laboratories
C. S. ANUEKWnN
Secretary, Board of tditurs
Previously unpublished papers appearing in this book may be reprinted,

and to
abstracted or abridged, provided credit is given to RCA REVIEW to the issue
tt.e author, or authors, of the papers in question. Reference
date or number is desirable.
Permission to quote other papers should be obtained from the publica-
tions to which credited.
NBC STUDIOS 6A AND 6B
BY
GEORGE M. NIXON
National Broadcasting Company. New York
Summary -The NBC Studios 6A and 6B, located at Radio City, New
York, were placed in operation early in November, 1941. The article de-
scribes the salient acoustical, electrical and architectural features of these
studios. The photographs supplement the text in illustrating the more
important details.
TWO new auditorium -type broadcast studios were placed in oper-

ation by the National Broadcasting Company at Radio City, New
York, early in November 1941. These studios, designated as
Studios 6A and 6B, incorporate the accumulated years of operating
experience in the careful planning, constructing, and decorating of
studios to coordinate properly all of the various operating requirements
of network broadcasting.
The two studios are similar in size and modern architectural design,
but distinctively different in decorative treatment to obtain a pleasant
contrast in appearance. The overall length of each studio is approxi-
mately 100 feet ; the width an average of 48 feet ; and the ceiling height
ranges from 13 to 191/2 feet. The stage section is 37 feet deep and 45
feet wide; and its ceiling height is from 13 to 16 feet.
Figures 1 and 2 show the overall appearance of one of the studios
Studio 6A. Figure 1 as viewed from the rear of the elevated and
-
stepped seating section shows the control booth, to the right of the
stage, from which the occupants have an unobstructed view of the stage
and of the major portion of the studio seating section. The opening
for the clients' observation booth window may be seen in the upper
right. From this booth the sponsors of the program may watch and
listen to the progress of the program. Figure 2 shows the studio as
viewed from the apron of the stage with the seats in the fore part of
the studio removed. The opening of the spotlight booth is seen in the
upper righthand corner of the picture.
The walls of Studio 6A and the stage ceiling are of a rich warm
copper color. The chairs in this studio are of dark -green fabric uphol-
stery, and the flooring has a rubber covering upon which the carpeting
is laid. The walls of Studio 6B and the stage ceiling are of a bright
cheerful silver color. This studio has a red rubber flooring, with red
carpeting, and blue upholstered chairs.
259
260 RCA REVIEW
Each studio accommodates an audience of about 450 in comfortably

upholstered chairs which are arranged to provide an unobstructed
view of the stage. One third of these chairs, 150, are removable and
are located on the level floor area immediately in front of the stage.
The remainder, 300, are fixed and arranged on a gradually stepped
slope so that all seats are "good seats." The seats of the chairs are
automatically retractable so that, when unoccupied, the seat slides
downward and backward to provide a maximum of space between
adjacent rows.
Fig. 1
The upholstered chairs, in addition to being comfortable, perform

the important function of maintaining substantially the same acoustical
conditions with and without an audience present in the studio. Pro-
gram balance and microphone location may be determined and adjusted
carefully during rehearsal, with the assuring knowledge that it will
not be disturbed by the influx of the audience. The removable seats in
the front of the stage permit use of this area by performing groups or,
by the removal of two or three rows of seats and the use of platforms
abutting the stage apron, the size of the stage may be augmented to
accommodate very large performing groups.
The wall and ceiling surfaces of the stage are arranged to provide
sound diffusion so that the reflected sounds are mixed properly and
NBC STUDIOS 6A AND 6B 261
the tonal quality of the performer or performing group is enhanced.

The wall on the control booth side is plane, but the opposite wall is
deeply "veed" in flat sections of wall, each 4 feet wide, to a depth of
2 feet, from the "crest" to the "trough" of the "vee" so that impinging
sounds are reflected dispersely. The rear wall is constructed of a series
of plaster semi -cylindrical convexly curved surfaces which provide
additional diffuse reflection. In Studio 6A the rear wall is of vertical
semi -cylindrical pillars and in Studio 6B these curved surfaces are in
horizontal layers. A drapery is provided which may be drawn to cover
Fig. 2
these surfaces to alter acoustical conditions in the studio, as desired.

Two other draperies are in each studio. One, for use as a "draw cur-
tain" in the front of the stage or what would be the proscenium in a
theater. The second is about midway along the side walls to reduce
the apparent size of the stage when very small performing groups use
the studio. The arrangement of the wall and ceiling surfaces may be
seen in Figure 3.
The stage ceiling, shown in Figure 4, is in "broken sawtooth"
fashion for two reasons, the first to provide a sound diffusing condition,
and the second to conceal the border lights and "spots" from the eyes of
the audience. This photograph, which is of Studio 6B, also shows the
horizontal semi -cylindrical surfaces on the rear wall of the stage; a
262 RCA REVIEW
closer view of the "veed" side wall and the method of concealing the
draperies in "curtain pockets ".
The acoustical treatment employed throughout the studio is rock
wool 2 inches thick except for the rear portions of the stage ceiling
where the rock -wool treatment is 4 inches thick. The acoustical treat-
ment on the walls is covered with perforated asbestos board applied in
2 foot squares. The treatment on the stage ceiling is covered with
perforated metal to provide greater absorption at the higher audio !!
frequencies.
The ceiling of the auditorium section is untreated except in rela-
tively small sections near the front of the stage to permit use of this
area for broadcasting purposes. The side walls of the auditorium are
Fig. 3
likewise untreated except near the stage to permit microphone usage

in this vicinity. The side walls are "splayed" horizontally in wide flat
panels so that very few opposite parallel surfaces exist. The absence
of acoustical treatment on the side walls and ceiling of the auditorium
section provide beneficial reflections from these surfaces which aid in
the transmission of sound from the stage to the listeners.
The rear wall of the auditorium is entirely treated with rock wool
2 inches thick covered with perforated asbestos board; in Studio GA,
the supporting surface is deeply serrated or "veed" and in Studio 6B,
horizontally splayed. Impinging sounds are largely absorbed by the
rear wall of these studios; the small amount of energy reflected is

diffused so that no annoying discrete, delayed reflections or "echoes"
from this wall interfere with the enjoyment of listeners in the front
seats.
The upholstered seats and broadloom carpeting lined with felt on
the aisles also provide a highly absorbent area in this portion of the
studio.
Sound isolation is provided by the use of double 6 -inch solid- cinder
block partitions to avoid sound transmission between the studio and
adjacent spaces. The walls, floor, and ceiling of the stage section are
supported on metallic springs damped by felt to reduce the trans-
mission of vibration between the studio and the building structure.
Fig. 4
All necessary openings in the boundary surfaces have been given

careful attention to insure maintenance of the sound isolation. The con-
trol booth and clients' booth windows are double glazed of % -inch and
% -inch thick glass set in felt. The window frame between panes is
lined with absorbent felt to decrease sound transmission and lessen
any tendency of resonance due to the confined air volume. Entrance
to the studios is effected through double doors separated by an acousti-
cally treated. vestibule. The doors are of solid wood 2% inches thick,
provided with sponge- rubber gaskets on had and sides, and have
264 RCA REVIEW
an automatic door closer of sponge rubber to seal the gap at the sill.
The air -conditioning ductwork is double wrapped with paper-covered
rock -wool blanket at all points where studio partitions are pierced.
The ducts are provided with cellular sound absorbers of Cabot's quilt
(ell grass) for a distance of 16 feet. This material effectively attenu-
ates sound that may travel through and along the ductwork.
The lighting system is not the least of the important features of
these studios. The seating section is pleasantly illuminated by cove
lighting in sweeping curves across the width of the studio. The light-
Fig. 5
ing forms an interesting pattern on the ceiling and provides indirect

illumination in this portion of the studio. The stage lighting uses
border lights with associated circuits arranged so that various color
combinations may be obtained to provide an atmosphere in keeping
with the program content. Spotlights are provided for individual
artists or featured small groups on the program. Individual control
of lighting sections together with master and sub -master controls
permit flexible operation to obtain desired combinations. The audi-
torium lights are dimmed at the start of the program to heighten the
enjoyment of the program by the visible audience. The stage lighting,
border lights, and spots may be seen in Figure_ .6. The observation
window of the electrician's booth is also seen in this photograph, and
Fig. 11
266 RCA REVIEW
through it some of the controls are visible. The general auditorium

cove -lighting system is best seen in Figures 1 and 2.
A low level of illumination is maintained in the control and client's
booths with concentrated "down" lights in the ceiling during the course
of a broadcast program. These lights, in the control booth, illuminate
the mixing equipment manipulated by the studio engineer and the
table, adjacent to the engineer's console, at which the producer sits
checking the script and the timing of the program. Auxiliary lights
are provided to increase the general illumination for rehearsals or main-
tenance of equipment. The lighting in the client's booth facilitates
Fig. 8
concentration on the stage because of the contrast in illumination. The

reduced booth illumintin during program time tends to reduce eye-
strain and eye fatigue for the occupants, and improves vision from the
control booth into the studio. A further aid to vision is the tilting of
the window so that light reflections from within the control booth are
reflected in such a manner that they are not readily visible to the
occupants of the control booth, particularly the studio engineer and
producer.
The broadcast equipment is "RCA all the way ". The controls are
mounted on a sloping panel set in a blond mahogany console. (The
console is shown in Figure 6.) The amplifying equipment is easily
accessible and located in a room just off the control booth, as shown
in Figure 7.
The sound -reinforcing system for the studio is energized from the
broadcast program circuit to permit the visitors to hear those artists
who perform close to the microphone and who would be otherwise
inaudible to a large portion of the audience. Another important reason
for the use of a sound -reinforcing system is that the program is bal-
anced for the microphone and all efforts are directed toward optimum
broadcasting results. Consequently, in view of the arrangement of the
'
orchestra, the choral group, sound effects, actors, etc., on the stage, the

f ,. f .i}'. .
t. i
,_
}
-}
i;` : ,,
If 'c .. t
e t _ :
.. ta.t
4. D
1
f"-t`-
r
4 -; 'a. t`
..!t .
1 ?I`f t v
1,. f '
% t !.
t j
t
t f ff 1 .'t .,
-ft -4 !. ti It
IT. it...
, Ir.
t . y
'At
-+ t.! s..#.
p
1 1
.
f.. ,1
lt< . .-_ -- t t t
. .. . :.. . .
o..
f
:
Fig. 9
program as heard directly in th studio from the seating area would

differ radically from that heard 'by. the radio listener. The use of a
sound -reinforcing system thereby provides the visible audience with the
same program balance as heard by the radio listener. Volume control
of the sound-reinforcing system is also provided at a small console
located in the seating section and shown in Figure 8.
An interesting view of the studio taken during a program from
mid -stage is shown in Figure 9.
The experience obtained from the operation of the studios should
further the knowledge of studio design and provide a sound basis for
future broadcast studio planning and construction.
268 RCA REVIEW
The planning of these studios was under the direction of Mr. O. B.

Hanson, Vice President and Chief Engineer of the National Broadcast-
ing Company. The architectural treatment, decoration, and general
construction details were the concern of W. A. Clarke, Manager of
Technical Services. The design and supervision of the installation of
the broadcast and electrical equipment were under the care of Mr. C. A.
Rackey, Audio Facilities Engineer and his associates, J. Hastings and
T. Nolen. The Operating and Architectural Departments of Rockefeller
Center were responsible for the preparation of plans and general super -
vision of construction.
GENERATION AND DETECTION OF FREQUENCY -
MODULATED WAVES
BY
STUART WM. SEELEY, CHARLES N. KIMBALL, ALLEN A. BARCO
RCA License Laboratory, New York, N. Y.
Summary -This paper contains a description of a frequency detector

which is inherently linear in operation and, consequently, adaptable for pre-
cision monitoring measurements. In conjunction with a mixer and hetero-
dyne oscillator, the detector is capable of operation at signal frequencies
encountered in the usual FM transmitter. The monitor detector is discussed
in the first part of the article.
The availability of a detector capable of demodulating frequency -modu-
lated waves with the introduction of less than 0.1 per cent distortion makes
possible the investigation of other low -distortion frequency-modulated
devices as, for example, a generator of phase or frequency- modulated waves.
The second part of the paper contains a description of a method of produc-
ing low -distortion frequency-modulated signals. The system was developed,
and its low -distortion capabilities realized, with the aid of the aforemen-
tioned linear detector circuit, in conjunction with an audio -frequency wave
analyzer.
PART I-LINEAR FREQUENCY -MODULATED MONITOR DETECTOR

PRINCIPLE OF OPERATION
INTRODUCTION
OPERATION of the detector unit may best be described with the

aid of the basic circuit diagram of Figure 1. ein is, in practice,
a frequency-modulated signal whose carrier frequency lies in
the range between 100 and 300 kc. The amplitude of e,, is sufficient
to overswing the limits of cut-off and zero bias of the 807 tube. During
portions of the cycle when this tube is cut off, the plate potential rises
to the +B level. During positive grid portions of any cycle the plate
current rises to a certain maximum value, beyond which it does not
increase with further increase in positive grid swing. (See plate family
curves of beam pentode tube characteristics.) Thus, the minimum value
as well as the maximum value of the instantaneous plate potential is
constant, and the output wave is squared off on top and bottom at
definite fixed potentials, in spite of possible variations in amplitude of
grid swing.
With a square wave of plate voltage, having a repetition rate equal
to that of the input signal eti,,, the action is as follows: when the plate
potential of the 807 reaches its peak value (equal to En) the small
condenser C (25 f approx.) in Figure 1 has charged through diode
di to a potential equal to EB, with the diode end of C negative. At that
269
270 RCA REVIEW
part of the cycle when the plate potential of the 807 swings toward its
lowest value, C discharges through diode d2 in series with its load Ro;
thus, one pulse of current flows through the load R of diode d2 for
each cycle of operation.
The total charge acquired by the condenser C once each cycle is
CE (neglecting the contact potential of diode d1). The portion of this
total charge which passes through diode d2 and Ro once each cycle is
equal to that total charge (CEe) minus the residual charge Cep, ti
;
(again, neglecting the contact potential of diode d2). Then, as long as
ER and ep, remain constant and the time constants of the charge and
discharge circuits are sufficiently small compared to the period of the
input wave, the total quantity of electricity which flows through Ro in
each cycle is constant. Thus, an increase in the repetition rate will
increase the number of current pulses per unit of time and, therefore,
Fig. 1
increase the average value of current flowing through Ro. Conversely,

a decrease in repetition rate decreases the average current through Ro,
and the average potential across Ro is thus a perfectly linear function
of the repetition rate or frequency of the input signal.
Obviously, then, dynamic operation of the detector will result in
essentially distortionless detection over that part of the characteristic
for which the previously discussed charge and discharge time-constant
requirements hold.
PRACTICAL DETECTOR CIRCUIT
The circuit of the laboratory model of the monitor detector, shown

in Figure 2, is essentially the same as the basic circuit of Figure 1,
except that a 6AG7 amplifier stage has been added.
Certain pertinent factors relating to the development and use of
the monitor detector are listed below.
(1) The sensitivity (audio volts developed per kc frequency devia-
tion in the input wave) is approximately a linear. function of the B
voltage applied to the 807. Hence, variations in this B voltage will
GENERATION AND DETECTION OF F-M WAVES 271
introduce undesired components into the detector output. Therefore,

the B supply must be well regulated.
(2) Approximately 20 peak -peak volts, at a center frequency of
150 kc, are required as input to the 6AG7, when the circuit of Figure 2
is used. Voltage at this frequency is obtained from that at any other
signal frequency by means of the heterodyne oscillator and converter,
shown in Figure 3. This particular circuit is designed for operation at
signal frequencies in the 40 -Mc range.
The a -c peak-to -peak grid voltage applied to the 807 must be at least
20 per cent greater than the value required to swing from zero bias
to cut -off. The 3000 -ohm series grid resistor in the 807 circuit aids
in maintaining approximate equality of duration of the two halves of
the plate -voltage wave, and also limits the peak grid current on positive
grid swings.
ztorrf
tsoir.+F
F-1
! F. INPUT
APPROA /So MC
47 70v P-P
Tt.--f
Fig. 2 -90 -volt and 150 -volt supplies are regulated. One side of all
heaters (supplied from 6.3 -volt transformer) is grounded.
(3) The 807 plate load is adjusted to produce plate-current satura-
tion at a peak grid signal just short of grid current.
(4) The value of capacitance C (in conjunction with other inherent
circuit capacitances) is determined by the maximum permissible time
constants in the charge and discharge circuits. These should both be
sufficiently small compared to a half-period at the highest operating
frequency to permit (a) ep to rise to within 0.1 per cent of EB in one -
half cycle and (b) to permit ep to fall to within 0.1 per cent of its
normal minimum value, again within a half- cycle.
(5) The lower limit of operating frequency is determined by the
cut-off of the low -pass filter in shunt to R (Figure 1). This filter is
necessary both to remove undesired signal- frequency components from
the output of the device and to allow the instantaneous potentials across
Ro to rise and fall with each discharge- curret pulse. If Ra were by-
passed directly, to remove the signal-frequency components, the residual
charge on the bypass condenser would act as a diode bias, and prevent
complete discharge of condenser C. Then the discharge currents would
272 RCA REVIEW
approach zero asymptotically as the frequency was increased, and the

resultant curvature of the output voltage vs. input frequency would
produce undesirable distortion.
The characteristic impedance of the filter is high compared to Ro.
The filter is terminated at its output and where it may be connected
to the grid of an audio amplifying tube.
(6) It has been found experimentally that a small bias voltage
(approximately 1.5 volts) in series with the charge diode d1 effectively
overcomes the contact potential of both diodes and contributes some-
what to improved linearity of the device.
To DETECTOb'
20r P-
AT APP.VO.I
/S0 Ac
.NIW T
i To Sr
PP
.T Si6 FRO.
150 r
_ PS," TONED TO 2"."
V NAAMON/C Of or APE?.
a [Z
C, dSO V f
Fig. 3
EXPERIMENTAL DETERMINATION OF DETECTOR LINEARITY

One may expect representative linear performance from a monitor
detector built according to the data presented. There may, however,
be circumstances wherein an accurate knowledge of the capabilities
of the device are required. For this reason a brief description of the
calibration method employed is included here.
Figure 4 shows a circuit in which the monitor detector (without
the heterodyne oscillator of Figure 3) is represented diagrammatically
by a box with input and output terminals. A signal generator is
employed (in conjunction with a calibrating crystal oscillator) to
provide a point-to -point determination of d -c detector output vs. input
frequency. An L. & N. Kohlrausch slide wire, fed from a 2 -volt storage
cell, is used in conjunction with a sensitive galvanometer in a poten-
tiometer circuit which permits the d-c output of the detector to be
,
GENERATION AND DETECTION OF F -M WAVES 273
determined with a precision of about 0.02 per cent (approximate read-

ing accuracy of the graduated scale of the slide wire) .
The extreme sensitivity of the d -c measuring circuit requires the
elimination of all sources of diode -current fluctuation. Hence, the B
supplies feeding the 807 plate and 6AG7 screen are doubly regulated
(for calibration purposes only) , with the circuit shown in Figure 4,
and the diode (d1) bias obtained by bleeder current (as shown in Fig-
ure 2) is replaced by a 1.5 -volt battery for determination of detector
linearity.
The crystal oscillator of Figure 4 is used to check the calibration
of the signal generator, by zero beat methods, with the detecting
phones in the plate of the mixer tube. Calibration of the signal gen-
MON/TOP DE TECTOP C4 ,BP4T,ON C /.PCU /T
GAL rANOYMErEi
MON /TOR
PREQUENCY
DETECTOR
?r ST0 461
CELL
L6N
kW/LORI/SCR
41 ,01 me(
/S0 Y
.DOUBLY PECULATEo
R7N/EP S!/OPL Y
P4yNES
SO r DOUOr vE6
300r 0PoM 0r /IS ULATfO SuPPUEo
PM/LAW ro iAG7 SCPfFN
SUPPL v -
47 Sr
7 ANO 407 ALATI
LOUN/:
N MON//OF Of,
r
GS7
ONE SLOE LK NEATEDS AVE S/Of Of HEATERS
000NNOEO OL1puNOEO
Fig. 4
erator at frequencies which correspond to fractional harmonics of the
crystal is possible as a result of curvature in the mixer characteristics.
Measurements of d-c output voltage of the discharge diode in the
detector circuit were made at regularly spaced input-frequency inter-
vals. Any non -linearity in the plot of d-c volts vs. frequency was
determined by plotting the readings, and comparing the resultant
curve with a straight line having approximately the same mean slope.
Then a curve of the difference in ordinates of the measured curve
and the comparable straight line was again plotted against frequency.
By employing this method, it was found that the departure from
linearity of the composite detector characteristic was less than 0.02
per cent over a frequency range of 50 to 250 kc. Thus, if a frequency -
modulated wave of 100 -kc deviation were applied to the detector at
a carrier frequency of 150 kc, the detector would introduce only
274 RCA REVIEW
approximately 0.02 per cent distortion in the audio output obtained
from the detector.
Tests were also made to determine the susceptibility of the detector
to amplitude modulation in the input signal. With 25 per cent ampli-
tude modulation (and a mean value of input signal of 20 volts peak to
peak) the output contains an undesired signal corresponding to less
than 0.05 per cent of the audio voltage produced by 100 -kc deviation.
The sensitivity of the detector is approximately one volt peak -to-
peak (audio) for 100 -kc deviation in the input signal (and 20 volts
peak -to-peak applied signal). Therefore, if a single frequency signal
of 100 kc is applied to the detector the d -c output should be approxi-
mately 0.5 volt. Knowledge of this sensitivity makes possible the
adjustment of the heterodyne oscillator frequency (in Figure 3) to a
value which causes the beat frequency to lie in the band for which the
detector linearity is optimum. It is necessary merely to measure the
d -c output of the detector as the heterodyne oscillator frequency is
varied, and to adjust the latter to produce a d -c output voltage corre-
sponding to 150 -kc carrier input.
APPLICATIONS
The monitor detector as thus constituted is an excellent instrument
for studying the effects of the pass characteristics of a frequency-
modulation receiver's r -f and i -f circuits on distortion, for checking
the performance of frequency-modulation signal generators, for use as
a monitor and peak deviation indicator at a frequency-modulation
transmitter, and for many other applications.
PART II -LOW DISTORTION FREQUENCY -MODULATION GENERATOR
INTRODUCTION
Frequency -modulation signals can be derived from systems based
on either frequency or phase modulation principles, provided that the
audio potentials (or program material) be properly integrated before
application to the phase modulator.
In a frequency -modulation system of the type employing a reac-
tance tube modulator, the frequency deviation produced by a certain
audio voltage is independent of the frequency of that audio voltage,
and is determined only by the sensitivity of the reactance tube (kc
deviation per audio volt), and by the subsequent frequency multiplica-
tion. Assuming that the circuits in the generator are sufficiently flat
over the desired deviation band, the harmonic content of the output of
a frequency -modulation generator of the aforementioned type is deter-
mined principally by curvature in the reactance tube characteristic,

and is not a function of audio frequency.
Conditions obtaining in phase -modulation systems are quite differ-
ent, since, for a given frequency deviation, the required peak -phase
deviation varies inversely with audio frequency, i.e.
57.3 X freq. dev. c/s at carrier freq.
c15 in degrees at carrier freq. =
audio -freq. c/s
Thus, the audio voltage at the phase -modulator grid must vary
inversely with audio frequency, to produce constant frequency devia-
tion. If the distortion in the resultant frequency -modulated signal is
caused by a non -linear relationship between the instantaneous phase
of the generated signal and the instantaneous audio voltage, the dis-
tortion will generally vary in some inverse manner with audio fre-
quency. Another source of distortion lies in the inability of the
frequency multiplier circuits to pass all the required side components
of the modulated signal. This deficiency will generally produce dis-
tortion increasing with audio frequency.
The distortion due to non -linearity in the phase characteristic can
be reduced by employing frequency multiplication from a relatively
low- frequency source of phase -modulated signals, to reduce the required
phase deviation at the modulator. Side -band clipping distortion effects
are suppressed by employing multiplier circuits having adequate band
width.
PRINCIPLE OF OPERATION OF EXPERIMENTAL

PHASE- MODULATION GENERATOR
In most generators employing phase modulation as a means for
ultimately producing frequency-modulated signals, modulation by the
audio components or program material is accomplished at a low value
of signal frequency (generally about 100 kc), which is subsequently
multiplied and heterodyned to produce the desired carrier frequency.
One system of phase modulation employed experimentally requires
first the generation of amplitude -modulated signals at a low signal
frequency, after which the carrier is shifted in phase with respect to
the amplitude- modulated sidebands, to produce a phase -modulated sig-
nal. The inherently remanent amplitude modulation present in this
developed signal is then removed by limiter action.
Another system of phase modulation is described in Patent No.
2050067, issued August 4, 1936 to Dr. Walter Van B. Roberts. This
system makes use of an unique characteristic of a simple parallel tuned
276 RCA REVIEW
circuit operated at 1 /x/2 times its resonant frequency. An investiga-

tion of the operation of this circuit together with following frequency
multipliers was made with the aid of the afore -described frequency-
modulation monitor. The basic principle of operation of the circuit
is as follows A parallel resonant circuit, tuned to V2 times the fre-
:
quency of an unmodulated low- frequency carrier (generally about 100

kc) is used as the plate load in a pentode whose grid is fed with voltage
at the source frequency (100 kc). See Figure 5. The phase of the
plate voltage of the pentode is then varied by changes in the series
resistance of the resonant circuit. Figure 6 shows the manner of
variation of 4 with R; between the limits of R = 0 and R = oo the total
phase change is 180 . One of the important features of the system
is that the absolute value of the plate-load impedance is constant and
a TO MUL T/PL /CR

C//.9/N
TIMED TO Ir,?
' TIMES SOURCE
FREQUE/YC Y
Fig. 5
equal in magnitude to the reactance of the circuit capacitance, for all

values of R from zero to infinity. Hence, no amplitude modulation is
present, regardless of the extent or amount of phase modulation.
Appendix I contains a development of the equations of phase vs. R,
and also a proof of the constancy of plate -load impedance as R is varied.
In practice, R in Figure 5 is the dynamic plate impedance of a
tube, whose control grid is supplied with audio-frequency potentials.
Since the variation of with R is non-linear, as shown by Figure 6,
a linear relationship between the instantaneous phase of the pentode's
plate voltage and the instantaneous value of modulator grid voltage is
obtained only over an interval for which a curve of Eg vs. Rp (for the
modulator tube) has the same shape and curvature of that of Figure 6.
The problems associated with matching the curves of 4 vs. R and Rp
vs. Eg will be discussed in greater detail later.
The phase -modulated voltage (at a carrier frequency of 100 kc) is
fed from the tuned circuit of Figure 5 through a chain of harmonic
multipliers to develop a frequency -modulated or phase-modulated signal
at the desired carrier frequency. The phase deviation is, of course,

multiplied by the same factor as is the center frequency.
The phase deviation required at the modulator is determined by the
desired phase deviation and by the order of frequency multiplication.
Thus, an expression for modulator phase deviation (at the 100 kc level)
for any audio frequency is
(1)O
57.3 X freq. deviation at ultimate carrier freq.
audio frequency X order of frequency multiplication
It is desirable to maintain the maximum required phase deviation
in the modulator circuit at as low a value as possible, to minimize non-
SO
.:r.-._
.,..r-'-';
;;,=a . ',';6$!ff._.
41449 'ii#gsWUMhik' i!if'S'4EfW1.4:tid5!Ylli
EE Ed :
PHASE /ST
CHARACTER /C
ro
CF C'/OCN, 7 of F/6 S
JO - . . ... ... _
'r.i_VEEEEMEMNEEE+..
ME
4.
JO
m -i:EEiiiL2EIiElviBE8a
M_Mgpppvvv
p
S5gbY:-:_:
qp yl
EER 0tl11m's
EEYgggg!
ERNE:::
APPROX.OCAr/ON OF
PREFERRED OPERAT/N6 PO/N7
IJ
\
11ENEi
MEEE,
IL 11111C= E
70 0 = 90
47 P = ^O
EEMENEM E
0 '-PR, ''.:'.`,:.:'V [,... fi
O
t..
i :10
3
Pm C
Fig. 6
linear effects at low audio frequencies (which require the greatest

phase deviation for a given frequency deviation). Two methods of
minimizing this low -frequency-modulator phase deviation are available.
The first involves starting at a reduced value of signal frequency
(lower than 100 kc) to obtain 2 or 3 times additional frequency multi-
plication, thereby reducing the required maximum modulator phase
deviation by a factor of 2 or 3. The limiting low value of source fre-
quency must be greater than the highest audio-modulating frequency.
A better way to reduce the low audio -frequency phase shift required
in the modulator is to employ a heterodyne action at some point in the
multiplier chain, by which the phase-modulated signal is heterodyned
from say, 1500 kc, down to a lower frequency, after which additional
multiplication is applied. This permits the realization of practically
278 RCA REVIEW
unlimited multiplication of the phase deviation without resorting to

very low source signal frequencies in the modulator circuit.
The heterodyne system was employed in the laboratory generator,
as shown in the block diagram, of Figure 7. The source of signals is
a crystal oscillator operating at 100 kc. This is fed through a filter
to a 6SJ7 with the tuned circuit of Figure 5 as its plate load.
The grid of the 6L6 modulator tube (triode connected) is supplied
with audio potentials, and its plate resistance constitutes the variable
R of Figure 5. The resultant phase- modulated signal is multiplied
successively by factors of 3 and 5 to produce a 1500 -kc voltage, whose
phase deviation is 15 times that effected by the modulator. In a
/00 KC
.b/Ii
1500 MC
OLyw(LNM(/
4500 Mc
(P/PL (P
/00 MC
OP/YCP
MOD
n/ee
CO NPY// 0.
M/lfN GSM/
Of ./0 !
A.OiO
SO(1PC[
Fig. 7
parallel chain of multipliers, an unmodulated 100 -kc signal (from the

same crystal oscillator) is multiplied by factors of 4 and 4, to produce
a 1600 -kc voltage, with no phase deviation. The 1600 -kc signal serves
as a constant- frequency heterodyning source, converting the phase
modulated 1500 -kc signal to the difference frequency of 100 kc, which
is taken from the mixer plate through a low-pass filter. The phase
deviation in this 100 -kc signal is increased over the modulator circuit
deviation by a factor of 15. Hence, the apparent sensitivity of the
modulator (degrees phase shift /audio voltage) is increased by 15 to 1,
allowing a reduction of non -linear distortion products (at low values
of audio frequency) by a factor greater than 15 to 1.
The 100 -kc heterodyne signal is then fed through a subsequent
chain of multipliers which raises the carrier frequency to 40.5 Mc,
with a phase deviation of 405 X 15 = 6075 times that produced at the
modulator. For distortion measurements this output signal is applied
to the converter of Figure 3, which feeds into the linear monitor
detector.
Figure 8 is a diagram of the phase -modulator circuit employed in

the laboratory generator. The component circuits indicated by the
blocks in Figure 7 are, in general, self-explanatory, but a few remarks
determined from development experience are in order.
The coupling transformers for the 300 kc, 400 kc, 1500 kc, and
1600 kc stages are double -tuned transformers employing inductive
coupling. The coupling and damping of each unit was adjusted to
provide adequate band -pass characteristics. In this respect it must be
noted that those stages following the heterodyne mixer require greater
percentage band width than those preceding the mixer by a factor
equal to the heterodyne ratio (15 in this case).
APPFO
7100.rr`
r0 MU[ 7/P[ /C I/
srAoEs
7NNE0 ro IT /00 EC
1_
TruNEO ru
/OO MC
"ION q
OfC
riNEO ro
o
! 1
T 5..( iS00 A
roo c
ONE S.DE OE 11EA7ERS O.oOUNOEO
B 275 Y
Fig. 8 -Phase modulator circuit.

The 4.5 Mc, 13.5 Mc, and 40.5 Mc stages utilize single resonant
circuits, each tuned with a 3 -30 f trimmer and all loaded or damped
sufficiently to make the pass band essentially flat and thus to minimize
the generation of amplitude modulation and phase distortion in these
stages.
Jacks are provided in each stage to facilitate the measurement of
d-c grid current, as a means for determining correct tuning conditions
in preceding stages. The grid bias resistors are adjusted to produce
maximum harmonic output in each stage. Thus, for example, in a
tripler whose plate load is tuned to 300 kc, the tripling action is
maximized by adjusting the tripler's grid leak to produce maximum
grid current in the following tube. This bias adjustment is carried
out in all stages. The grid- circuit time constants of all stages are
made sufficiently low to permit the multiplier grids to follow any a -f
variations in the amplitude of the signals, thereby producing limiting
action.
280 RCA REVIEW
EXPERIMENTAL PROCEDURE
An essentially distortionless audio source is required for measure-
ments of phase distortion in the frequency-modulation generator. This
happens because the amplitude of the nth harmonic in the developed
frequency deviation is proportional not only to the amplitude of the
nth harmonic in the audio-source voltage, but also to the order of the
harmonic, since the frequency deviation is proportional to the rate of
phase deviation. Thus, 1 per cent 3rd harmonic in the audio input
would produce (in an otherwise distortionless system) 3 per cent har-
monic in the resultant frequency deviation, which would then appear
as 3 per cent 3rd harmonic in the audio output of the monitor detector.
This problem may be circumvented in two ways, i.e., first integrate
the audio voltage before application to the modulator, in which case
the device produces frequency- modulated signals, requiring constant
audio input for constant -frequency deviation as the audio frequency is
varied. The source harmonics are then reduced in magnitude by an
amount proportional to their order. The alternative method was used
in the experimental work; an audio source of good waveform was used
in conjunction with a series of band -pass filters, each capable of passing
an octave of frequencies, starting at 30 cycles and ending at 8000
cycles. No integration was employed : hence, the generator produced
phase -modulated signals.
A General Radio Type 636 -A wave analyzer was used for all dis-
tortion measurements. The harmonic content of the audio source was
found to be less than 0.05 per cent for all audio frequencies employed.
The wave analyzer was then connected permanently across the monitor-
detector output terminals. In parallel with this a 9 -inch oscilloscope
was connected to facilitate preliminary adjustments. In general the
audio input to the modulator was adjusted to produce 100 -kc deviation
(for all audio frequencies) at the final carrier frequency of 40.5 Mc.
The heterodyne oscillator in the detector unit was adjusted to produce
150 kc as the intermediate frequency for application to the monitor
detector. This had previously been ascertained as lying in the region
of optimum detector linearity.
As pointed out in a preceding section, there are two sources of
distortion, in a generator of this type; the first occurs at low audio
frequencies (less than 400 cycles) and is due to non-linear relation-
ships in the modulator, which are accentuated by the large audio
swings required.
This type of distortion is generally a function of the modulator
bias, which controls the degree of matching between the curves of
vs. R (Figure 6) and RD vs. ED of the modulator tube.
At higher audio frequencies (greater than 1000 cycles) the dis-

tortion is generally independent of modulator bias (since only a small
audio swing is applied), but is determined, rather, by sideband clipping
in the multiplier tuned circuits.
In adjusting the unit for minimum distortion products the audio
input was set at 3200 cycles, and, with the monitor detector output
indicating 100 -kc deviation, and the wave analyzer reading 2nd har-
monic distortion, adjustments were made by loading and tuning of the
multiplier -load circuits until the distortion due to sideband clipping
was minimized. This procedure was repeated at 7500 cycles. The over-
all distortion was reduced to less than 0.5 per cent rms, for 100 -kc
deviation, at these higher audio frequencies.
D/STO.OT/ON ,f, .8145
SLQ MODULATOR MBE
TR/ODE CONNECT/ON Ed r 125
TREQY/ENCy .DEV/AT/ON CONSTANT AT iqp eC
/TOR IO. sMC CA.VRIER J
MODl4 4TOR / OR/O AMS
Fig. 9
The audio frequency was then reduced to 400 cycles, for distortion
measurements at relatively larger modulator audio swings. With the
frequency deviation adjusted for approximately 100 kc and the wave
analyzer set for 2nd harmonic determinations, it was found that the
amplitude of this harmonic was a function of modulator bias, and that
a minimum value of less than 0.1 per cent could be attained with critical
modulator bias adjustments. This sharply defined minimum is due to
the dependence of the localized curvature of the RpEg curve of the
modulator upon its bias, and to the fact that, for a certain bias, correct
matching between this curve and that of 4. vs. R is obtained. Figure 9
shows curves of rms distortion vs. bias at 400 cycles modulating volt-
age.
At 50 cycles the applied audio voltage required for 100 -kc deviation
is 8 times that required at 400 cycles, and as a result, the degree of
matching of the Eg vs. Rp and 4 vs. R curve is less complete over an
282 RCA REVIEW
audio cycle. However, the rms distortion at 50 cycles is relatively

constant at 3.5 per cent over a 2 -volt range in bias, for 100 -kc devia-
tion, as shown in Figure 9.
A large number of tube types (6J7, 6K7, 6V6, 6AC7/1852, 6AC7/
1853, 6J5, 6L6) were used experimentally as modulators. The
6AC7/1852, for example, has higher sensitivity (kc deviation per
audio volt), but the dependence of distortion on bias is more critical.
The 6L6 was finally chosen as a result of its lower rms distortion
at 50 cycles modulating frequency, and the relative constancy of dis-
tortion with bias.
Analysis of the modulator circuit shows that distortion produced
at low audio frequencies, caused by mismatch of the Rp vs. Ep and
4 vs. R curves, is due principally to insufficient curvature in the Rp vs.
1J/STO9T/ON VS. FgEQUENCY DEY/AT/ON
6L6 MODULATOR TUBE
AUD/O FREOUENCY = 50 CYCLES
S ::::1
MODULATOR B/AS '?/. 7 V
y /o
4--
It
;_
f .fi
7rt
JJ
I
,:eCO!/ENCY
,
Fig. 10
DEV/AT/ON
t`
vC
l 11.7
E,, curve. This can be corrected, if desired, by some types of regen-

erative audio circuits, which tend to increase non- linearity in the
modulator tube Rp curve, or by the use of an additional modulator to
supply 2nd harmonic variations in Rp to produce additional curvature.
EXPERIMENTAL RESULTS
Figure 10 shows the distortion incurred at 50 cycles modulating
frequency vs. frequency deviation (in the 40.5 Mc carrier) taken at
a value of bias for which the 2nd harmonic distortion is at a broad
minimum.
Figure plot of rms distortion vs. audio frequency for two
11 is a
values of frequency deviation. Note that the distortion is essentially
independent of deviation except at frequencies below 200 cycles, for
which the audio swings (approximately 20 peak-peak volts at 50 cycles)
are large, causing considerable mismatch in the phase and Rp curves,
which, at lower swings, are normally coincident. .
GENERATION AND DETECTION OF' F -M WAVES 283
0 3uti.
/00 kC
DE V/A 7/ON OVEAACL .D/STO.OT/ON vs Au0/O
FREQUENCY
,i 6L6 MODULATOR i

FIRM
a9fV/A7/ON \ URI= .:
.: :? m
..
a
SO /00 ;GOO ,0000
AUD/O FQS QuENCY - CYCLES PER SECOND
Fig. 11
The distortion at high audio frequencies is independent of modu-
lator bias, but is very much a function of tuning, even with sufficient
circuit damping. This is particularly the case in the 300 -kc tripler
following the mixer stage, which must accommodate a band of fre-
quencies 15 times as great as that encountered in circuits directly
following the modulator.
Attention must be directed toward minimizing amplitude modula-
tion in the frequency-modulated wave, particularly at the lower signal
frequencies, for which the phase deviation is small. As shown in
Appendix II, n per cent amplitude modulation can produce n per cent
2nd harmonic distortion in the frequency deviation, due to sideband
cancellations, if the phase deviation is sufficiently low to permit the
representation of the wave by a carrier and a pair of first -order
sidebands.
APPENDIX I
.4: - s,,, cry,
lOOKc)
-
Analysis of Phase Modulator Load Circuit:
CAM. /7-
To /47.4
TuYED
KC
Fig. (a)
(1) To demonstrate the constancy of tuned impedance as R is
varied. to, = 2,r X input frequency (100 kc)
1 1
(R + jLco,) (RCw, + j Lao 84)
jCa, jCt),
R+j L(v,-
Cw,
1
) RC w, + j (LC wa2 - 1)
284 RCA REVIEW
The circuit is resonant at V2 X the source frequency, i.e., C in the

above figure is half the value required for circuit resonance at the
1
source frequency. Therefore, LC =
20)82
and (FtC0), 0).2 1 1
2w,2 / je,8C 2RCw, -1- j1 1

zZ =
RC,), + j
w, a 2RCw,' - '1
% C
1
21,z
The absolute magnitude of 73 is therefore 'z: 1
which is in-
0),C
dependent of R.
(2) The phase vs. R equation of the circuit is readily determined

from the above, viz:
(2RCoa j1 1 1 1- j 2RCw,
2 RC,o, - j1 jw,C w,C 2RC(,), - j1
total phase shift 4) = phase shift in numerator 0 minus phase shift in
denominator a
c=0-a
-1
4) = tan-1(- 2RC(08) - tan-1
2RCw,
tan 4) =
tan O - tan .a
1 + tan tan a O
Hence,
1
-2RCw. +
2RCw. 1 1
tan 4) = 2RCco)
2RCw. 2 2RCw.
1 +
2RCw.
1
= tan-1 RCw8
4RCw,
This is plotted in Figure 6
APPENDIX II
Proof that amplitude modulation in a low- deviation phase-modulated
wave can produce frequency -modulation distortion equal approximately
to the per cent amplitude modulation, if this factor is small.
First assume that the phase deviation is sufficiently low to justify
the use of only the first -order sidebands. The signal then has the
form of Figure (b).
Fig. (b)
= 4)m sin (Oat
With R added amplitude modulation
2E,
= tan-1 sin (oat
E,
where E.= phase-modulation sideband amplitude
E,= carrier -voltage amplitude
= 27r X modulating frequency
Now let the carrier be amplitude modulated at the same audio fre-
quency, viz :
e, =E0(1 +m sin (oat)
The phase angle is given by
2E, sin wet
= tan-1
E,(1 + m sin teat)
2E.
Let be a constant, determined by the amount of phase moclu-
Q.
lation.
sin coat
Then 4 = tan -1 K
1 + m sin coot
For small angles it is permissible to write
sin mat
43=K
1 + m sin (Oat
286 RCA REVIEW
also for m, (per cent amplitude modulation 100) less than 0.10, it
is sufficiently accurate (within 1 per cent) to write 4 = K sin wat (1
m sin wat) = K sin wat -
Km sine wat. This can be expanded to give
-
= K (sin wat - --m + -m
2 2
cos 2wat)
The ratio of coefficients of the 2nd harmonic and fundamental audio

m
terms is m, 2, which implies the existence of -X
100 per cent 2nd
2
harmonic in the phase-modulated wave.
The frequency deviation is equal to

1 4
27r dt
Performing this operation we get
dt
= K (wa cos wat - 2w;,, -
ni
2
sin twat)
Of
4
=- -=Kfa
1
(cos wat -m sin 2wat)
dt
27r
where .1f is the frequency deviation (instantaneous) and fa is the
modulating frequency. It is seen that m per cent amplitude modulation
will therefore produce ni per cent 2nd harmonic distortion in the fre-
quency deviation of the system.
A NEW CHEMICAL METHOD OF REDUCING THE
REFLECTANCE OF GLASS
BY
F. H. NICOLL
Research Laboratories, RCA Manufacturing Company, Inc., Camden, N..1.
Summary-A new chemical method of reducing the reflectance of glass

is described. It is compared experimentally with previously known chemical
methods of reducing the reflection, and is shown to be superior in many
respects. The new method produces a tough hard film of very low reflecting
power. The treatment involves exposure to hydrofluoric acid vapor and is
applicable to large sheets of glass. The process requires neither vacuum
nor expensive equipment and is suitable for many optical glasses. A number
of possible uses of nonreflecting glass produced by this method are men-
tioned. Photographs are given of several examples of these applications,
including cathode -ray -tube faces, ground-glass screens on cameras, and glass
covers for photographs and pictures.
INTRODUCTION
THE use of an externally applied transparent film on glass, for

the purpose of reducing the reflection of light from its surface,
has been known for some time.'," Considerable practical use
has been made of evaporated films one -quarter of a wavelength thick.'
These films have been used to increase the transmission of many optical
devices employing lenses, prisms, etc., with very satisfactory results.
Many obvious applications of such films have not, however, been
realized, due either to lack of hardness of the film or to the high cost
of vacuum evaporation.
In striving for improved contrast on television cathode -ray tubes,
it has been observed that the contrast of the picture is always greater
when the picture is viewed in the dark. This is partly due to the scat-
tering of extraneous light from the fluorescent screen, but there is also
the effect of specular reflections from the surface of the glass. Such
reflections of illuminated surroundings noticeably reduce the contrast.
In the case of monitors and oscilloscopes used on outdoor equipment,
specular reflection of the observer himself may cause considerable
annoyance. The elimination of such extraneous reflections is very
desirable. With this and other applications in mind, an investigation
has been carried out of chemical methods for reducing these surface
reflections.
287
288 RCA REVIEW
EXPERIMENTS ON THE OLD CHEMICAL METHODS OF TREATING GLASS
Chemical methods of treating glass to reduce its reflection have

been known for a very long time. Early workers such as Taylor,5
Kollmorgan,6 and Wright? have observed and investigated the effects
of a number of chemical solutions and the films which they produced.
Later Vasicek8 measured the indices of refraction of such films and
showed them to be quartz. Similarly, Blodgett3 came to the conclusion
that the film was quartz. She obtained reflections of 21 per cent and
10
o .2 1.4 I.G 1.8 20

INDEX OF REFRACTION OF FILM.
Fig. 1- Theoretical reflection from glass coated with a transparent
film % wavelength thick, in per cent of incident light.
30 per cent of the original reflection by chemically treating glasses

having refractive indices of 1.72 and 1.61, respectively. Recently she
has applied the technique to the making of gauges') for measuring the
thickness of various monolayers. During the preparation of this paper,
Jones and Homer10 have published the results of some further experi-
ments on the early chemical methods of etching lead glass.
The conditions for extinguishing reflection of monochromatic light
have been given very fully by Blodgett$ for the case of glass coated
with a thin transparent film. If multiple reflections are neglected, and
light of normal incidence only is considered, then for films of optical
REDUCING THE REFLECTANCE OF GLASS 289
thickness one -quarter of a wavelength viewed in air of refractive index

1, the resultant reflected intensity is given by
la= (B -C)2
where B = (1-n1) (1 + n1)
and C = (n, - ng)/ (n, -F ng)
The quantities ?If and ng are the refractive indices of the film and
glass respectively. The incident light intensity in this case is taken as
unity. The resultant reflection IR has been calculated for various
/O
II `
r
8
fo 0
z
o
h
J
LL
W
a
114
f
2. I. f /.6 /2 2.0
INDEX OF REFRACTION OF FILM
Fig. 2- Theoretical reflection from glass coated with a transparent film
1/a wavelength thick, in per cent of original reflection.
glasses and films of different indices of refraction. The curves are
plotted in Figure 1 where the reflection is given in per cent of the
incident light. They extend from values of n,. = ng down to the value
of n1 corresponding to minimum reflection.
These curves are useful for comparing the results for different
glasses. When one glass only is involved it is more convenient to have
the reflected intensity in per cent of the reflection from the untreated
glass. For this reason the curves have been replotted in Figure 2.
With these curves it is possible to estimate the index of refraction of
the film from the intensity of the reflected light. When the film thick-
290 RCA REVIEW
ness has been adjusted to give minimum reflection on a piece of glass

of known index of refraction, then the intensity of the reflected light
gives a measure of the index of refraction of the film. This is most
easily obtained from Figure 2 since the simple photocell method of
measurement lends itself more readily to a comparison of the intensi-
ties of the reflected light from the treated and untreated glass, than
to absolute measurements.
Preliminary experiments on the chemical method of producing
transparent films were made using a number of inorganic solutions
such as were employed by the early workers. Samples of various optical
glasses were immersed in these liquids for varying lengths of time.
The reflection was compared with that from the untreated glass using
a photocell and a tungsten light source for the measurements. No filter
was used, since experiments indicated that it was not necessary for an
approximation to the sensitivity curve of the eye. It was found that
with many solutions the reflection of flint glass of refractive index
about 1.6 was reduced to about 25 or 30 per cent by prolonged treatment
at 80 degrees Centigrade. Crown glass of index about 1.52 was more
difficult to affect and the reflection was never reduced below about 70
per cent of its original value and in most cases was unaffected.
Referring to Figure 2 we see that the results on both flint and crown
glass correspond to a film having a refractive index between 1.4 and
1.5. This is in agreement with the assumption of other workers that
the film was quartz. Because a greater percentage reduction was
desired, it was clear that a heavier flint glass would have to be used.
Since X -ray protection glass has an index of about 1.7 and has a very
high lead content, it appeared to be very suitable. The high lead con-
tent is necessary since the portion of glass removed by the old chemical
processes is mainly lead oxide.
Various inorganic solutions, such as were used by some of the early
experimenters, were tried for their action on X -ray glass. It was
immediately found that the majority of the solutions produced tough
transparent films showing interference colors.
The amount of chemical action, and, hence, the thickness of the film,
was found to be a function of time, temperature, and concentration.
A limited range of these variables was investigated and no attempt
was made to correlate them all in a systematic manner. The measured
reflection for various film thicknesses was, however, obtained as a func-
tion of time. Figure 3 shows this reflection in per cent of the reflection
from untreated glass, as a function of the time of immersion in a one
per cent solution of sodium acid phosphate at 50 degrees Centigrade.
The process caused the formation of a white film of soluble lead salts
on the surface of the glass, in addition to the low reflection film of
quartz.
9
The reflection measurements and estimate of color were made after

this white deposit had been washed off. It will be noticed that the
second -order minimum corresponding to a thickness of 3/4 wavelength
has been reached and passed. The colors of the film observed by
reflected light are noted along the time scale. These indicate what
colors are not being reflected. For instance, a purple film means red
and blue are being reflected while green and yellow are not. A film
of this color gives a minimum reflection for light most sensitive to the
eye. The minimum of 14 per cent reflection fits in fairly well with the
result expected from Figure 2 for a film of quartz. The film was very
hard and could be rubbed off only with a rough lap. It could be com-
pletely removed by a quick dip in boiling NaOH. This process was,
in fact, an excellent method of cleaning the samples before an experi-
ment in order to get consistent results.
The time required to produce a purple film of minimum reflection
is also a function of temperature. Figure 4 shows the approximate
relationship for X -ray glass. It can be seen that the time is reduced
by a factor of nearly ten as the temperature is increased from room
temperature to 80 degrees Centigrade.
A number of additional inorganic solutions were also tried, all of
which were effective to a greater or less degree. The table below gives
some of the solutions used and the time necessary to produce a purple
film.
TABLE I
Efficiencies of various solutions in producing a purple
film of minimum reflection on X -ray protection glass.
Time of Exposure
in Min. Solution Concentration Temperature C
7 Fel (SO4) i 1% 2G
2 (SO4):1 1% 80
50 NH.,PO4 1% 27
15 NaH:PO4 1% 50
9 NaH:PO4 1% G3
5 NaH..PO4 1% 80
45 H;;PO4 1% 80
No effect H;;PO4 Conc. 80
100 K_,Cr.:07 2% 2G
2 i{..Cr:07 2% 80
5 CuSO4 2% 80
No effect
in producing
interference
colors NaOH 2% 27
292 RCA REVIEW
These chemical films, on being left exposed for a period of days,

gradually increased in reflecting power, sometimes as much as two
or three times. However, it was found that heating restored them to
nearly their original values. This was thought to be due to the
absorption and subsequent removal of water vapor. A similar effect
has been noted in a recently published paper,10 in which the authors
mention that baking stabilizes the film.
There seems to be a very limited use for this type of chemical film
where a reduction in reflecting power is desired. The fact that the
films are very hard is an advantage but, on the other hand, the process
is applicable only to the heavier flint glasses and the reduction in
reflection is limited by the index of the film. This is a serious limita-
/00
z
o
_
J W
50 Z
Cc'
W J W W
J 7 CWl
o
,T
o
o 20 4-0 60 80 ioo izo
TIME OF In n]ERSIoIV-- MINUTES.
Fig. 3-Reflection as a function of time of immersion for X -ray

protection glass in 1 per cent NaH2PO4 at 50 C.
tion for applications such as cover glasses for meters, viewing windows,
etc. Because of the higher index of refraction of X -ray glass, the
initial reflection is about 1.6 times that of window glass, as can be
seen from Figure 1. It follows that even if the reflection of X-ray
glass is reduced to about 15 per cent, this only corresponds to a
reduction of window glass reflection to about 24 per cent. This in
itself is not low enough for most purposes. In addition, the cost of
X -ray glass is sufficiently high to prohibit its use for some applications.
For most applications requiring non -reflecting glass, a method of
treating glasses of low refractive index is required. At the same time
the method should be applicable to large pieces such as sheets of win-
dow glass. Further experiments were therefore undertaken in the
hope of finding a chemical method which would satisfy these require-
ments.
A NEW METHOD OF PRODUCING FILMS OF LOW REFLECTING

POWER ON GLASS
As long ago as 1900, Rayleigh11 pointed out in one of his papers

that hydrofluoric acid diluted one part in 200 of water removed a
thickness of glass corresponding to about 1/4 wavelength of light per
hour. At the same time the glass, if agitated remained perfectly
polished. The thickness corresponding to 1/4 wavelength is of par-
ticular interest in the production of nonreflecting films. If some
compounds of low refractive index were left on the glass while the
remainder of the glass was dissolved away, then a dilute solution, such
as Rayleigh used, could produce a film of low reflecting power.
60
Ill
el
40
0 2.0
aim
40 (00 60
Teriv c.
Fig. 4 -Time of formation of purple films of minimum reflection as a
function of the temperature of the solution, for X -ray
protection glass in 1 per cent NaH2PO4.
With a view to investigating the effect of hydrofluoric acid liquid

and vapor at the same time, some samples of window glass were placed
half in the solution and half out. A solution of one part concentrated
hydrofluoric acid (48 per cent) to 200 parts of water (1/4 per cent HF
solution) was used, the solution being contained in a 50 cc waxed
beaker with a waxed glass lid. After 64 hours at room temperature,
the glass in the liquid had been thinned appreciably, but was still
polished in appearance. Immediately above the liquid, however, were
interference colors ranging from blue near the liquid to straw color
about 2 or 3 mm above the liquid. These colored films appeared to
be perfectly transparent and the glass underneath did not appear to
be etched or pitted by the acid vapors.
This result was sufficiently encouraging to justify further investi-
gation. Experiments were therefore continued using 1, 2, and 4 per
294 RCA REVIEW
cent HF solutions with the sample suspended horizontally over the

liquid. The results were now more uniform and the films could be
tested and examined more readily, since the area was much larger. It
was observed that condensation of the water solution of HF on the
film during its formation removed the film or made it very fragile.
For this reason, it was found desirable to cool the bottom of the con-
tainer to a temperature below that of the glass in order to prevent
condensation. Water from the faucet at about 17 degrees Centigrade,
about 10 degrees below room temperature, was found to be satisfac-
tory.
The most satisfactory arrangement for producing uniform films
seemed to consist of a shallow tray two or three inches high with the
/0 .V
TIME OF E.:POSURZ - HOURS
Fig. 5- Reflection as a function of time of exposure for window
glass in the vapor of per cent HF.
1
flat piece of glass to be treated placed over the top, its lower face
exposed to the vapor of the acid. The tray itself was supported by
blocks about 1/I inch high. Water to a depth of about 1/2 inch was
circulated around and under the tray at a temperature of about 10
degrees below that of the room. The glass was usually cleaned with
powdered chalk and water, although this was not always necessary.
Under these conditions, with a 1 per cent solution, about 6 to 10
hours (depending on temperature) was required to produce a film of
purple color on a piece of ordinary window glass. This film reduced
visible reflections to a minimum, about 6 to 10 per cent of that from
untreated glass.
Figure 5 shows a curve of reflection against time of exposure for
one side of a piece of glass exposed on a tray 2 inches deep. The
water bath was kept at about 15 degrees Centigrade and the average
room temperature was 25 degrees Centigrade. The glass used was
7 REDUCING THE REFLECTANCE OF GLASS 295
1/2-inch Libbey Owens Ford "Double Strength" window glass, glazing

quality. The acid solution was 1 per cent HF and about 50 cubic
centimeters were used, covering the bottom of the tray to about 1/16
inch. Some of the products of the etching remain as a white film on
the glass, but these are subsequently washed off with water. The
reflection measurements for this curve were made with this white film
remaining on the glass, but the approximate values of the reflection
after the film was washed off are given by the broken line. The curve
of Figure 5, like that of Figure 3, extends through the first-order
minimum to the second -order minimum. The colors obtained corre-
sponded to those given in Figure 2, but they were more vivid since
the refractive index of the film was more nearly that required for
extinction of the reflection at the minimum.
Some investigation was made of the effect of concentration and
there was evidence that about 5 per cent HF was the greatest con-
centration that would produce a suitable hard film without heavy
etching. The more concentrated the solution, the more rapid the pro-
duction of the film. For instance, the vapor from the 48 per cent HF
produced a purple film in about two minutes at room temperature, but
the glass was so heavily etched that it became translucent.
These films produced by hydrofluoric acid vapor, were proved to
be true interference films of high light transmission by comparing the
transmission of glass with and without a film on it. The results were
similar to those with evaporated transparent films. The transmission
was considerably increased at the same time that the reflection was
decreased. A pile of nine plates of window glass had a measured trans-
mission of 51 per cent, which is quite close to the theoretical value.
A similar pile of nine plates of treated window glass with the reflection
at each face reduced to between 8 and 10 per cent gave a measured
transmission of about 88 per cent, which is within a few per cent of
the calculated value if the absorption in the glass is taken into account.
In addition to the increase in transmission, an image viewed through
the treated plates had much better contrast, and was practically free
of multiple images. These results show that the transparency of the
films produced is very high and that very little etching takes place.
THE BEHAVIOR OF VARIOUS GLASSES UNDER HYDROFLUORIC

ACID TREATMENT
A number of different types of glass were tested for their reaction
to hydrofluoric acid vapor. In a number of cases non -reflecting films
were produced and the results, where possible, were correlated with
the composition of the glass. Libbey Owens Ford window glass
produced very fine nonreflecting films. In particular, the "Single
296 RCA REVIEW
Strength" and "Double Strength" types were satisfactory. Plate glass

was usually satisfactory, but in some cases spurious marks were
brought up, although no evidence of polishing marks was seen.
Several of the glasses forming good films by the above technique
were known to contain more than 10 per cent calcium oxide. On the
other hand, some of those not forming a film were known to contain
less than 5 per cent calcium oxide. This led to the belief that the
films were calcium fluoride formed by the action of the hydrofluoric
acid vapor, the other products being removed as vapors or during
washing in water. This view is borne out by reference to Figure 2.
Glass with an index of about 1.52 has its reflection reduced to about
6 per cent at the minimum. This is seen to indicate a film of index
slightly less than 1.3. Calcium fluoride is one of the few possible
substances that could be formed from the glass and possess such a
low index of refraction, unless a skeleton film is formed. This latter
possibility seems to be doubtful since a drop of oil could be wiped off
the film without affecting the reflection or color of the film.
A considerable number of small samples of various optical glasses
were also tested. The compositions were not known, but the index of
refraction n and the approximate values of the dispersion (np no) -
of the various samples are given below. nF and ne are respectively the
indices of refraction for the F and C lines of the solar spectrum. The
glasses are tabulated in three columns according to the ease with
which the hydrofluoric acid vapor produced a nonreflecting film. Those
listed as good formed hard films of low reflecting power with little
or no etching. Those listed as fair tended to etch visibly and in some
cases the films were only faintly colored, indicating a film of too high
an index of refraction. The samples listed as poor usually produced
heavy etching and no interference films, while in some cases only a
faintly colored brownish film could be obtained.
TABLE II
Quality of Films Produced on Various Glasses
Good Fair Poor

n (nr - nc) u (nip- nc) n
1.5147 1.5232 1.5123
1.5158 1.619 .0169 1.525
1.5159 .0095 1.6214 .0172 1.7 (X -ray glass)
1.5166 .0088 1.47 (Pyrex)
1.5177
1.5179
1.5243 .0085
1.6204
1.6214 .0170
The compositions of the various glasses in the above table were

not known, but it is highly probable that a number of the optical
glasses contained at least 10 per cent calcium oxide. On the other
hand, X -ray glass which formed no interference film contained no
calcium oxide. These results were sufficient to show that quite a wide
range of optical glasses can be treated by the HF vapor process. In
particular it seems to be applicable to those glasses which do not have
their reflections reduced appreciably by the old chemical methods. The
process, therefore, extends very greatly the usefulness of the chemical
methods.
PROPERTIES OF FILMS PRODUCED BY HYDROFLUORIC ACID VAPOR
Several properties of this type of nonreflective film have been

noted and in a number of cases they have been compared with those
of an evaporated film of a satisfactory type. Some chemical solubilities
have been tabulated below for convenience. Solubility and insolubility
are indicated by the letters S and I.
TABLE III
Chemical Properties of HF- Produced Films and Evaporated Films
Solution HF- Produced Evaporated

Film Film
Dilute HF S S
Boiling NaOH S S
(NH4)2CO3 sol'n I S
Dil. H2SO4 I S
Cold Dil. HNO3 I S
Cold NaOH I S
Conc. HC1 I S
Cold H2SO4, Conc. I S
Boiling H2SO4, Conc. I S
Cold Chromic Acid I S
Boiling Chromic Acid I S
From this table it can be seen that the chemically produced non-
reflecting film is very much more resistant to chemical action than
the evaporated film. The solubility of the chemically produced film in
dilute HF is quite a useful feature in some respects. If it is desired
to remove the film from one portion of a glass sample, it is possible
to do this and leave a sharp boundary by merely dipping the sample in
8 to 10 per cent HF for a few seconds. The film can also be removed
in any desired pattern by swabbing with this solution.
298 RCA REVIEW
In the case of physical properties the difference between these two

types of film is also important. The chemically produced film was much
harder and resisted rubbing better than the evaporated films. It could
Fig. 6- Photograph of framed picture with glass front, one -half

of which is nonreflecting.
also be cleaned with water, alcohol, or some commercial window clean-
ing solutions. However, rubbing with powdered chalk, such as is
present in some window cleaners, gradually removed the films. Both
types of film withstood heating to red heat in air. The HF-produced
Fig. 7- Experimental cathode -ray tube face, one -half of which

is nonreflecting both inside and out.-
film on a piece of window glass was exposed to air in a furnace at 560

degrees Centigrade for about an hour, as a test of the ability of the
film to withstand the temperature necessary to seal a glass window on
the cone of a cathode -ray tube. This treatment altered the reflection of
this particular sample by only 10 per cent.
Compared with the silica film produced by the old chemical method,
the HF- produced film is somewhat less hard. However, the absolute
reflection from one surface of a piece of window glass treated with
HF may be as little as 0.25 per cent of the incident light, whereas the
silica film produced on X -ray protection glass reflects about 0.94 per
cent of the incident light. The HF- treated glass is therefore a marked
improvement over the other type for purposes where a low reflecting
power is desired. It should be pointed out here that these nonreflective
films, like any other types, tend to increase their reflection as they
accumulate dust. Occasional cleaning overcomes this difficulty. Simi-
larly, finger prints are very noticeable since the presence of the oil
increases the reflection considerably. It is best to wipe these off with
alcohol or some similar solvent.
USES OF NON -REFLECTIVE FILMS PRODUCED BY

HYDROFLUORIC ACID VAPOR
The first use that suggests itself for these nonreflecting films is
that of making lens surfaces nonreflecting, and of higher light trans-
mission, as is now being done with evaporated films. Its use here, how-
ever, is limited by the fact that all optical glasses cannot be treated
with hydrofluoric acid vapor to produce nonreflecting films. At the
same time those which can be treated will take different times, depend-
ing on their composition. Nevertheless, its application to lenses should
not be out of the question if suitable glasses are used.
This type of film can also be used to make meter fronts non-
reflecting, as was demonstrated by Blodgett3 with built -up films. These
HF- produced nonreflecting films also gave a marked improvement in
this case, particularly under glare conditions.
Store show windows may also be made nonreflecting by the HF
vapor process since it is applicable to large sheets. However, no tests
of its effectiveness in this application were made. On the other hand,
photographs, etc., covered with nonreflecting glass had their appear-
ance improved by the reduction of extraneous reflections. Figure 6
shows a framed picture, one -half of which has a nonreflective film on
the glass front. In this case the reflection was about 10 per cent of
that from the untreated glass.
Figure 7 shows the result of treating one -half of the face of an
experimental cathode-ray tube. The face of the tube had the non -
reflective film removed from one -half, but the other half was non -
reflective on both inside and outside faces. The glass front on this
tube was waxed on the cone of the cathode -ray tube. The photograph
was taken with the tube face illuminated by diffuse light. On the
one -half of the tube face, the nonreflective film has practically elimi-
300 RCA REVIEW
nated the reflections, and the contrast of the scanned test pattern is
much improved.
These fluoride films have also been tried on the ground-glass focus-
ing screens used in cameras. As usually used, the unground side of
the glass is exposed, and reflections from this surface interfere con-
siderably in focusing a dim image on the screen. A nonreflecting film
reduces this specular reflection so that it is practically invisible. This
improves the contrast and makes it easier to focus accurately. A
photograph of such a treated ground glass camera screen is shown
in Figure 8. It was taken with the screen illuminated by diffuse light
coming from the left. The lower half is seen to give a much clearer
Fig. 8-Ground -glass focusing screen on a camera, one -half of

which is nonreflecting.
image due to the presence of the nonreflecting film. At the left -hand
edge of the upper half of the screen the image is also clear. This is
because the light shield surrounding the screen prevented the diffuse
light from reaching the ground glass. This is, therefore, the best image
obtainable, since it corresponds to no stray illumination on the screen.
It can be seen that the nonreflecting film produces practically the
same result.
Similarly, a nonreflecting film has been found to improve the con-
trast on a ground -glass screen used for back -projected pictures such
i
as those used for home movies. A sample screen was made about six
inches square. An improvement was noted when an image was observed
under ordinary room illumination with the unground surface treated,
and a reduction in milkyness of the ground surface was noted when
the ground side was treated.
It
BIBLIOGRAPHY
1J. Strong, "On a Method of Decreasing the Reflection From Non -

Metallic Surfaces," J.O.S.A., Vol. 26, p. 73, 1936.
2 C. H. Cartwright and A. F. Turner, "Reducing the Reflection From
Glass by Evaporated Films," Phys. Rev. 55, p. 595, 1939.
3 K. B. Blodgett, "Use of Interference to Extinguish Reflection of Light
from Glass," Phys. Rev. 55, p. 391, 1939.
4 A. F. Turner, "Coating Lenses," Phot. Technique, p. 48, October 1940.
H. D. Taylor, "The Adjustment and Testing of Telescope Objectives,"

T. Cook, York, England, 1896.
6 F.
Kollmorgen, "Light Transmission Through Telescopes," Trans.
Soc. Ill. Eng. II, p. 220, 1916.
7 F. E. Wright, "Characteristics of Optical Glass," Ordnance Depart-
ment Document, No. 2037, p. 76.
g A. Vasicek, "New Method for Investigating the Refractive Index and
Thickness of Interference Films on Glass," Phys. Rev. 58, p. 924, 1940.
9 K. B. Blodgett, "Silica Gauge for Measuring Thickness by Means of
Interference Colors," Rev. Sci. Inst. 12, p. 10, 1941.
10 F. L. Jones and H. J. Homer, "Chemical
Methods for Increasing the
Transparency of Glass Surfaces," J.O.S.A., 31, p. 34, 1941.
11 Rayleigh, "On Polish,"
Scientific Papers of Lord Rayleigh, Vol. 4,
p. 546.
4!
AN ANALYSIS OF THE SIGNAL -TO -NOISE RATIO OF

ULTRA- HIGH -FREQUENCY RECEIVERS
BY
E. W. HEROLD
Research Laboratories, RCA Manufacturing Company, Inc., Harrison, N. J.
Summary-This paper presents an elementary analysis of the effect of

the various sources of fluctuation noise on the signal -to -noise ratio of radio
receivers. Because the noise induced in negative grids at high frequencies is
included, the work is particularly applicable at ultra -high frequencies. It is
found that the signal-to -noise ratio depends on the antenna noise; in addi-
tion, when bandwidth is not a. consideration, it depends on the ratio of
equivalent noise resistance to input resistance of the first tube, and, when
bandwidth is a major consideration, on the product of input capacitance and
equivalent noise resistance. The coupling from antenna to first tube is an
important variable in receiver design and an optimum coupling is found
which results in an improvement in signal-to -noise ratio. This optimum
condition is often considerably different from the adjustment for maximum
gain and, by its use, the noise induced in the grid becomes relatively un-
important. The noise from the second stage of the receiver is also evaluated.
It is shown that the thermal noise from a wide -band interstage circuit may
be made negligible by concentrating all the damping on the secondary side.
Calculations of typical receiver arrangements using triode type 955 and pen-
tode type 954 mixers are given for 300, 500 and 1,000 megacycles.
I-INTRODUCTION
HE useful reception of radio and other types of signals is lim-
ited, in the main, to those signals which exceed the unavoidable
random fluctuation noise of the communicating system. Because
there are other sources of noise in addition to normal random noise,
such as static and man -made interferences, the signal must usually
exceed the random noise by a considerable factor, under most conditions
of reception. However, since random noise imposes a readily calculated
absolute limit to sensitivity, it is usually used to designate receiver
performance. This criterion is particularly appropriate when the noise
which is inherent in the receiving antenna is included. At low fre-
quencies, up to say 20 megacycles, the chief sources of noise in the
usual receivers are due to atmospherics and allied causes, to thermal
agitation in the antenna" and in the input circuits of the first tube,
1F. B. Llewellyn, "A Rapid Method of Estimating the Signal-to -Noise
Ratio of a High Gain Receiver", Proc. I.R.E., Vol. 19, pp. 416 -420, March,
1931.
2 R. E. Burgess, "Noise in
Receiving Aerial Systems ", Proc. Phys. Soc.,
Vol. 53, Part 3, pp. 293 -304, May 1, 1941.
302
SIGNAL-TO -NOISE RATIO OF U -H-F RECEIVERS 303
and to shot-effect in the plate circuit of the first tube. At higher fre-
quencies, however, the noise induced in the input grid by the passage
of electrons3.4.5 must also be considered as a contributing factor. When
the gain of the first tube is low, it is also important to consider noise
contributed by the coupling circuit to the following tube and also the
noise contributed by the following tube itself.
In all cases, the most convenient evaluation of the total noise energy
is made by referring all sources of noise to a given point in a receiver.
For example, if the grid of the first tube is used as reference point,
the noise voltage resulting from shot noise in the plate may be divided
by the gain of the tube to give an entirely fictitious, but equivalent
noise voltage at the grid reference point. This noise voltage may be
added in the usual mean -squared manner to other independent noise
voltages referred to this same point to give a total noise value. In
estimating receiver performance, however, it is necessary to refer the
received signal to this same point, if the signal -to -noise capability of
the receiver is desired. In a companion paper,' D. O. North has shown
that the absolute sensitivity of a receiving system depends upon a noise
factor, N, of the receiver itself. The relations found in the present
paper will show how N varies with tube and circuit constants and how
receivers may be designed to minimize it.
Studies of signal -to-noise ratio in the input circuit of receivers
have previously been made qualitatively by Llewellyn' and quantita-
tively by Williams' and by Frnz.8 The present paper is an extension
of the work of these men to include wide -band and ultra- high-frequency
applications. In addition, a study of interstage coupling has been
included for those cases when the gain of the first stage is low. This
information is particularly necessary in ultra-high-frequency super -
heterodynes when a wide band is used with a low-gain converter stage.
II -BASIC RELATIONS
A tuned antenna of radiation resistance Rd will be assumed to be
coupled to the receiver through a perfect transmission line of charac-
teristic impedance Zo = Rd. Coupling to the input of the first tube will
1 loc. cit.
8
Stuart Ballantine, "Schrot Effect in High- Frequency Circuits ", Journ.
of Frank. Inst., Vol. 206, pp. 159 -167, August, 1928.
4 D. O. North and W. R. Ferris, "Fluctuations Induced in Vacuum -Tube
Grids at High Frequencies ", Proc. I.R.E., Vol. 29, pp. 49 -50, February, 1941.
5
C. J. Bakker, "Fluctuations and Electron Inertia ", Physica, Vol. 8,
pp. 23 -43, January, 1941.
' D. O. North, "The Absolute Sensitivity of Radio Receivers ", RCA
REVIEW, January, 1942.
7 F. C. Williams, "Thermal Fluctuations in Complex Networks ", J.
I.E.E., Vol. 81, pp. 751 -760, December, 1937.
8 K. Frnz, "The Limiting Sensitivity in the Reception of Electric Waves
and Its Attainability ", Elektrische Nachrichten - Technik, Vol. 16, p. 92, 1939.
304 RCA REVIEW
be assumed to be through a network which is the equivalent of a trans-

former whose leakage reactances are eliminated by tuning, and whose
step -up, or effective turns ratio, is m *. An equivalent circuit then is
as shown in Figure 1. The total input impedance of the tube with the
antenna disconnected, as determined by circuit and lead losses as well
as transit -time loading, is lumped as R1 in the figure. The open -circuit
antenna signal voltage is shown as Ca. Feedback effects which may
change the signal-to -noise ratio in the tube will be neglected in this
analysis. The noise sources which must be considered are:
1. Thermal and other noise in the antenna.
2. Thermal noise of circuit and leads.
3. Induced input electrode noise in first tube.
4. Plate noise of the first tube referred to the input circuit.
5. Noise of parts of receiver subsequent to the first tube,
and also referred to the input circuit.
1
T
I:m
Fig. 1 -Basic equivalent circuit of receiver input.
The noise of the antenna is conveniently introduced by considering

this noise as if it were noise due to thermal agitation in the radiation
resistance of the antenna and assigning to this resistance an effective
temperature, Ta. Under ordinary circumstances, at frequencies below
20 megacycles, this effective temperature may be considerably higher
than the ambient temperature. At ultra -high frequencies, where
directive antennas are employed and atmospherics are negligible, the
value of Ta will more nearly approach the ambient temperature and
may even go below it. When a conventional method of measuring
receiver sensitivity is employed, a resistor at ambient temperature is
substituted for the antenna resistance and the results obtained will
correspond to Ta equal to the ambient temperature. These cases will
be given special attention in this paper.
To distinguish between items 2 and 3 of the above list it will be
convenient to divide R1 into two parallel components, one having
thermal noise and representing circuit and leads, and the other having
induced grid noise and representing electronic loading. If the con-
m is the ratio of output voltage to input voltage of. the network.
*
K. G. Jansky, "Minimum Noise Levels Obtained on Short -Wave Radio

Receiving Systems ", Proc. I.R.E., Vol. 25, pp. 1517 -1530, December, 1937.
SIGNAL -TO -NOISE RATIO OF U-H-F RECEIVERS 305
ductances of these two components are go and g,, respectively, then

i
gsl+ge =
R1
The noise induced in the input circuit by the passage of electrons
has been found by North and Ferris' for tubes with an input control
grid adjacent to the cathode. Their results for oxide -coated cathode
tubes showed the noise to be equivalent to the thermal noise of a re-
sistor of a value equal to the electronic loading and whose temperature
is about 5 times room temperature. Furthermore, their work has shown
that, theoretically, to a first approximation, induced grid noise may be
added to the plate noise (referred to the grid) of the same tube as if
they were independent sources of noise. Their results have made the
application of this analysis to ultra -high frequencies possible.
With respect to induced grid noise in tubes whose control grid is
not adjacent to the cathode, the problem is not yet completely solved,
although a start has been made by Bakker.5 The emphasis in the
present paper will be on the more common tubes with oxide -coated
cathodes and an adjacent control grid which is also the input grid.
However, for the sake of completeness, the results which apply to the
other forms of tube are given in an appendix.
One of the more common feedback effects in ordinary tubes at high
frequencies is due to cathode lead inductance which introduces an input
loading very similar in nature to the electronic loading. However, it
has been shown10 that this feedback leaves the signal-to -noise ratio
unaffected, at least in the first approximation. In using the results of
the present paper, therefore, input conductance due to cathode lead
inductance should not be included in the evaluation of the electronic
conductance, ge. If the conductance due to the cathode lead is in-
cluded in go as if it were an ohmic loss the error will be small in most
practical cases.
The plate noise of the first tube will be referred to the grid in the
usual manner by making use of an equivalent- noise -resistance concept.'1
In many practical cases, the gain of the first tube will be sufficiently
high so that noise of succeeding stages will be negligible. However, to
preserve complete generality, the other sources of noise will be assumed
to be included by adding a second equivalent noise resistance which is
' IOC. Cit.
5 IOC. Cit.
1 M.J. O. Strutt and A. Van der Ziel, "Methods for the Compensation
of the Effects of Shot Noise in Tubes and Associated Circuits ", Physica,
Vol. 8, pp. 1 -22, January 1941.
See Part V of B. J. Thompson, D. O. North and W. A. Harris,
"Fluctuations in Space- Charge- Limited Currents at Moderately High Fre-
quencies", RCA REVIEW, Vol. V, pp. 505 -524, April 1941 and Vol. VI, pp.
115-124, July 1941.
306 RCA REVIEW
calculated at the grid of the first tube by using the squares of the gains
between this point and the actual sources of noise, in the usual manner.
A later section will include the evaluation of this second equivalent -
noise- resistance factor as applied to the second stage. The total equiv-
alent noise resistance at the grid will be designated by Reg and will be
the sum of the noise resistance of the first tube and that calculated
from all succeeding stages.
Using the above concepts, by assuming an overall receiver passband
which is not wider than that of the circuits considered in the analysis,
it now becomes possible to redraw the circuit to include all the noise
sources. This is done in Figure 2 in which all noise sources are shown
as constant -current generators except the equivalent- noise -resistance
source, Reg which is shown as a voltage. Throughout this analysis, Ta
represents the effective antenna temperature, TR represents ambient
or room temperature in degrees Kelvin, k is Boltzmann's constant
e2 = 4k TR Rey of
1
1m
2= 4k Ta
Ra
=4k TRgnAf = 5 x 4k TRgenf
Fig. 2- Equivalent circuit of receiver input with noise sources included.
(equal to 1.37 X 10 -33 joules per K) and f is the overall effective

passband of the receiver for noise purposes." The induced noise in
the input grid is shown in the form suitable for oxide -coated cathode
tubes with control grid adjacent to the cathode.
Since the antenna resistance referred to the input circuit secondary
is equal to m2Ra (where m is the effective turns ratio, or step -up, of
the input circuit), the antenna noise referred to the grid is just the
thermal noise of a resistor m2Ra at a temperature Ta in parallel with
1
the resistance R1 = already there. Thus, another equivalent
ga -1- ge
circuit can be shown in Figure 3a.
In order to simplify the analysis still further, let us replace the
reflected antenna conductance, 1 /m2Ra, by the symbol g0. The two
resistances of Figure 3a in parallel will then be equivalent to the single
resistance of Figure 3b. Furthermore, by remembering that
11 Thompson, North and Harris, loc. cit.
SIGNAL-TO -NOISE RATIO OF U -H -F RECEIVERS 307
1
gn+ge=
we can rewrite the expression for the constant-current noise generator
as shown in the latter figure. It will be well to keep in mind that the
condition for maximum gain (impedances matched) corresponds to
gaR1 = 1.
e2 =4kTRRegpf
M/'
n
To
4k TRef
1
in2 =
TR m2 Ra + 9n + 5 9e
e2 =4kTR Regpf
To
in2
= 4k TR Lf 19a + R + 49eJ
(b)
Fig. 3-(a) Equivalent circuit with antenna loading and noise shown as
reflected values in secondary of antenna transformer; (b) Simplified equiva-
lent obtained by introducing ga for the reflected antenna conductance.
When the circuit is reduced to the simple one of 3b, the relations
for the signal voltage and noise voltage are readily written down. As
to the signal, in Figure 3b it could have been represented as a constant-
current generator of value ea /mRa. In terms of the reflected antenna
conductance, go, the signal voltage on the tube grid is then
gaRl R1
egrid = ea (1)
1 + ga R 1 Ra
This expression reduces to the expected value for matched impedances
if gal?, is given the value unity.
The mean -squared noise voltage on the grid, from Figure 3b, is then
308 RCA REVIEW
Ta 1 R12
41C T - (2)
R1 (1 + gaRi)'
The signal -to -noise ratio is given by dividing (1) by the square -root
pif (2) and is
erignal
_- - - - ea- X
k T r Ra pf
VI ITT
1 2
Ta Rea Rev 1 Rev (3)
+ 2 + (gaRl) -}- 1+ + 4goR1
TR R1 R1 (gaRi) Ri
'I'he denominator inside the radical of Equation (3) is the noise factor,
N, whose importance in limiting the absolute sensitivity is discussed
in a companion paper.'
For the sake of completeness, another relation, which is of some
interest, is the bandwidth of the input circuit (which has been assumed
equal to, or greater than, P f) . The total bandwidth, from points 3 db
down on each side of the resonance curve, may be compared with that
of a simple tuned circuit of capacitance C shunted by a resistance
R1
and is
1 + gaRi
1 + gaRi
Input band width = F (for 3 db down)
2irCR1
where F is equal to unity for an input circuit which is equivalent to a
single-tuned circuit. For coupled circuits, or other band -pass arrange-
ment, F may be somewhat greater.
Equations (1) to (4) are the fundamental relations which are
applicable to the input circuit. Their interpretation will be made more
clear by the subsequent discussion.
III-CONDITIONS WHEN GAIN IS MAXIMUM (IMPEDANCES MATCHED)
From Equation (1), maximizing with respect to gaRi, it is found

that the maximum grid signal is given when gaRi =1 as, of course, is
expected. This condition means simply that
maRa. = Ri
6 D. O. North, loc. cit.
SIGNAL -TO -NOISE RATIO OF U-H -F RECEIVERS 309
and the antenna impedance is exactly matched to the tube impedance

by the input circuit. It is the usual condition of adjustment and war-
rants further consideration.
The signal -to-noise ratio (from Equation (8) ), when g0R1 =1, is
ea 1
eefgnni
(5)
]g0R1=1 et2 Reg
v en: +1)-F 4 + 4 ge Ri
TR
where et2 is the open- circuit, mean-squared thermal noise voltage of

the antenna at room temperature. This expression shows that the
signal -to-noise ratio for a given signal, antenna noise and bandwidth
depends only on the ratio of the tube equivalent noise resistance* to
the total input resistance and on the fraction of this input resistance
which is electronic in nature. The latter fraction, shown as geRi in the
equation, must lie, of course, between 0 and 1. Thus, the signal-to -noise
ratio must always lie between the two limits imposed by the above
equation for ge = 0 and for geRi =1.
The minimum open-circuit antenna signal which will just equal the
noise under the matched -impedance condition, is given by
e
min
et2
V 2 + 4
Reg
R1
+ 4 geRi (6)
an expression which is obtained by setting Ta = TR as in receiver

measurements with a dummy antenna. This minimum value may be
used as a criterion for receiver performance.t Values of et' computed
for Ta = 293 K, Rd = 75 ohms, and various bandwidths are as follows:
of, et2,
ke V
10 0.11
200 0.49
2000 1.55
4000 2.2
* i.e., when noise of other tubes is small and Req is substantially due to
first tube, only. Otherwise, R, includes noise of following stages also.
As North has shown in Reference 6, ea /V et 2 cannot be changed by
antenna design except by changing the directivity.
310 RCA REVIEW
With the condition of matched impedances, which applies to Equation

(6), the signal, when measured at the receiver input terminals, will
be just half of the open- circuit signal, ea.
A curve of the factor (2 + 4 Req /Ri + 4 geRi) plotted against `
Ri /Req is shown in Figure 4. The lower curve is for the case when
geRi = 0 while the upper curve is for the case when geRi = 1. Results
for all practical receivers, when adjusted for maximum gain, must lie
between these two curves. Further discussion of these data will be
given later in the paper.
20 W
tp
24
z
a
z
18 Z
W
z
z Q
t.) 4 12 w
>
O
geRi=l
Q
2 6 tri
JW
_z
CO
u
w
.01 0 I O IO 100 0
RATIO OF INPUT RESISTANCE TO EQUIVALENT
NOISE RESISTANCE (Ri /Req)
Fig. 4 -The relative fluctuation noise of a receiver whose input circuit is

adjusted for maximum gain.
IV- CONDITIONS WHEN SIGNAL -TO -NOISE RATIO IS MAXIMUM
Referring to Equation (3) of the basic relations, if this expression

is maximized with respect to go, it will be found that best signal -to-
noise ratio is given when
R1
(gaRi)2 =1 + (1+ 4 geRi) (7)
Reg
In terms of the step -up, ni, of the input circuit the maximum signal -
to -noise ratio is given when
R1
r>t = 1 +- (1 + 4 geRi)
Ra
v Reo
Substituting (7) in (3) gives the maximum signal -to -noise ratio as
SIGNAL -TO-NOISE RATIO OF U-H -F RECEIVERS 311
eslerial ea
ea ,/ / Ta
+2
Req
-}-2
1
ReQ
1-
Req
+4geRt
N/ 2 maw t
/ TR R1 Rl R1
It is again seen that the signal -to-noise ratio for a given signal, antenna
noise, and bandwidth, depends only on the ratio Req/Ri and on geR,.
The minimum open -circuit antenna signal which will just equal the
noise is then
e.] min --V --et z

/ 1 + 2
ReQ
R1
+2
V
Reg
R1
1 +
Req
R1
+ 4 geRi (9)
w
20
1 1111 1
ADJUSTED FOR BEST

1111 ITT 24 z
1
SIG. /NOISE
10
-- ADJUSTED FOR MAXIMUM GAIN
8
6
.01 0.1 10 IO 100

NOISE RESISTANCE (R,/Req)
Fig. 5 -The relative fluctuation noise of a receiver whose input circuit is

adjusted for best signal -to -noise ratio as compared with one adjusted for
maximum gain.
where Ta has been set equal to TR for the lowest possible antenna noise
(as with a dummy antenna) . A curve of the factor included under the
right hand radical plotted against R, /Rcq is shown in Figure 5 for each
of the two limiting cases g(RI = 0 and gR1 = 1. In order to facilitate
comparison, the curves of Figure 4 are drawn with dotted lines.
It is, of course, necessary to sacrifice signal -voltage gain in order
to achieve the improved signal -to -noise ratio. The ratio of the signal -
voltage, antenna -to-grid gain (Equation (1)) with best signal -to -noise
ratio to that with optimum -gain coupling is
Actual Gain 2 g0R1
(10)
Max. Gain 1 + geR1
312 RCA REVIEW
where gaR, is the value given by Equation (7) for best signal -to-noise.
Equation (10) represents, therefore, the gain- reduction factor.
Curves showing the reduction in gain as well as the improvement
in signal-to -noise ratio"- which are made possible with this coupling in
comparison with optimum -gain coupling are shown in Figure 6. It is
seen that over the range of values usually encountered (Ri /ReQ < 10),
the signal-to -noise ratio improvement is appreciable though not
startling even for geR1 =1. The reduction in gain is very nearly equal
to the improvement in signal -to-noise ratio in this range. It should be
noted that the bandwidth of the overall input circuit is increased by a
change in coupling from that for best gain in the direction of that for
best signal -to -noise ratio. The amount of the increase is readily found
from Equation (4) using the value of gaR1 found from Equation (7).
s
I 0 1/
/\ .
,IrP/Ci
6
0
qeR
QJ
\ryPi"
'-
i ,.1
/'
////
\t,'"
OJGVEMENS
P MP N SN
\Q
R=O
I
\
I
0.01 01 I.0 10 100

NOISE RESISTANCE (R,/R,)
Fig. 6- Comparison of coupling for best signal -to-noise ratio, S /N, with
coupling for best gain.
V-DISCUSSION OF RESULTS OF SECTIONS III AND IV

Perhaps the most interesting feature of the analysis has been that,
no matter whether impedances are matched or whether best signal -to-
noise ratio is desired, the signal-to -noise ratio depends mainly on the
ratio of equivalent noise resistance, Reg, to the total effective shunt
resistance, R1, of the input to the tube. When Reg is chiefly due to the
first tube (as it is when the gain of the first tube is reasonably high)
and R1 is also chiefly a result of the first tube (i.e., when circuit losses
are negligible), a figure of merit for the first tube is Ri/ReQ. Unfor-
tunately, this ratio varies with frequency. Its use, however, together
with the curves of Figure 5, lead to a clear picture of receiver per-
formance. It is seen, for example, that in an ultra-high- frequency
* i.e., the reduction in minimum signal which it is possible to receive.
SIGNAL -TO-NOISE RATIO OF U-H-F RECEIVERS 313
amplifier or converter tube, an improvement in input resistance is just

as valuable as a reduction in equivalent -noise resistance.
Referring to Figure 4, for the impedance- matched condition, the
data show that little improvement is to be obtained in signal -to-noise
by an increase in Ri /ReQ much greater than unity. Thus, if impedance
matching is maintained, and a tube is found to have a minimum signal
factor corresponding to Ri /ReQ =1 it will not be possible to improve the
performance much more, unless perhaps geRi can be reduced by a
reduction in electronic loading so as to change from the upper curve
towards the lower. However, if a mis-match is permitted, as shown in
Figure 5, a definite improvement in signal-to -noise can be expected as
Ri/Req is increased even up to 100.
At all times, the minimum thermal noise voltage of the antenna
(Ta = TR) sets a definite lower limit to receiver sensitivity. As an
example, consider the minimum usable signal of a television receiver.
A receiver with a 4- megacycle bandwidth designed for the reception
of amplitude -modulated signals must have a signal -to-noise ratio of
around 30 db for a satisfactory picture. Since the antenna thermal
noise is 2.2 microvolts (for a 75 -ohm antenna), a satisfactory picture
requires a signal of 70 microvolts as a minimum and no amount of
improvement in the receiver can possibly decrease this value.
One of the conclusions to be drawn from Figures 4 and 5 is that
the presence of induced grid noise does not affect the signal -to-noise
performance by a large amount. Under most practical high-frequency
conditions, Ri /ReQ is well under 10 and if a tube and circuit whose
major loading is electronic is compared with one whose loading is
entirely resistive a maximum difference of only 1.4 db will be found
when the coupling is adjusted for best signal-to -noise ratio.
The curves of Figure 6 indicate that adjustment for maximum gain
(impedance-matching) is reasonable only as long as Ri /ReQ < 1. If the
latter ratio can be improved markedly, by an improved circuit, tube, or
both, so as to exceed unity by a significant amount it becomes profitable
to mis -match somewhat to take advantage of the improved signal -to-
noise ratio which is then attained.
There is still one other aspect of the signal-to -noise problem which
is of importance in the normal broadcast band, for example. In a
receiver design which is suitable for a wide variety of antennas, it is
customary to couple the antenna very loosely to the input circuit. In
this way variations of antenna impedance will have a minimum of
effect on receiver line-up and performance. The condition of loose
coupling corresponds to a very small reflected conductance, i.e.,
gaRi 1. This condition, when included in Equation (3), shows the
signal -to -noise ratio to be
314 RCA REVIEW
gaR1
e signal eG
Reel
2 V et'- 1+ + 4 geRl
/1/ R1
If gR1 is held fixed, the ratio is again seen to be of primary

Req /R1
importance. Greatest improvement in signal -to-noise, however, will be
accomplished by increasing the antenna coupling so as to increase ga.
VI -- SIGNAL -TO-NOISE RATIO WHEN BANDWIDTH IS OF

PARAMOUNT IMPORTANCE
This section is concerned with those instances where the unavoid-

able tube or circuit capacitance is so great that the input circuit band-
width is too narrow when the coupling is adjusted for best signal-to-
noise (the bandwidth, in this case, will be even narrower for the
impedance- matched condition). It will now be necessary either to
adjust the reflected antenna conductance for the correct bandwidth,
or to decrease the shunt resistance R1 by adding a loading resistor.
From Equation (4), it is seen that the relation between R1, the band-
width, the reflected conductance, and the capacitance is fixed as
2.7rp f' C R1
gaR1= 1
F
AZ' Rl
1
F
where Pw is introduced for 27rpf. The bandwidth Pf' is the r -f circuit
bandwidth and is to be distinguished from 0 f which is the total
effective bandwidth for noise purposes. When this expression for
gas, is substituted in Equation (3) it is found that the signal-to -noise
ratio is
1
-, /
esignal ea
(12)
Ta pwC Reg 1 + (puC Reg) /F + 4 gel?'
en2 V e=2 + +
TR F (Auk' R1) /F -1
It is now evident that, in this special case and once the r-f bandwidth
is assigned, a low value of Reg C is the desideratum for best signal -to-
noise. When Reg and C are mainly contributed by the first tube, a pos-
1
sible figure of merit for the first tube is then which is
2ir C Rea
expressible in cycles per second. It is also evident that, when R1
tilGN,11,-TO-.`'Ill Sh' RATIO OF ('-H-F RE('EI ['F'Kti 315
approaches infinity, the best possible upper limit to the signal -to -noise
ratio is obtained since the right hand term in the denominator then
vanishes. The input circuit bandwidth is then obtained by the antenna
loading only. Thus, the addition of a loading resistor to obtain band -
width always results in a needless reduction of signal -to -noise ratio.
A point of some interest is that a pair of coupled circuits, with
coupling adjusted for flat -top response and with R, approaching infinity
gives a value of F \12. With this coupled -circuit input arrangement,
the minimum signal which is just equal to the noise is then
'min = \/e " \' + 0.7127r1f) C R,g
1
where Ar is still defined by points 3 db down at each side of the

resonance curve.
VIL --- EFFECT OF OUTPUT CIRCUIT AND SECOND TUBE ON TOTAL

EQUIVALENT N oISE RESISTANCE, Rev
A. Maximum gain: bandwidth not important --In some receivers,

particularly those used at. ultra -high frequencies with a low -gain ampli-
fier or converter, it becomes necessary to consider the thermal noise
and tube noise subsequent to the input stage. The calculation is most
convenient if carried out in terms of an equivalent noise resistance
due to the second circuit and tube, but referred to the grid of the first
tub. The value so calculated must be added to the equivalent noise
resistance of this first tube itself to obtain the total Reg which has
been assumed in the preceding analyses. The results are applicable
whether the first tube be an amplifier or a converter. Induced grid
noise will he neglected in this section.
In receiver arrangements where the bandwidth of the circuit
coupling the first and second tubes is. a priori, wider than the overall
bandwidth of the receiver, the coupling circuit is designed for maxi-
mum gain. With a single tuned circuit of resonant impedance R2 and
with a second tube of equivalent noise resistance h' the mean -squared
noise at the grid of the second tube is proportional to .R_, + R,g.,l.
This may be referred to the grid of the first tube by dividing by the
square of the grid -to -grid gain. The total effective equivalent noise
resistance at the grid of the first tube is then
R., + Repz
R,ee = Rege + ( 13)
(GAIN) 1
where R,v is the equivalent noise resistance of the first tube. When
a hair of coupled circuits is used, with coupling adjusted for maximum
gain, it is necessary to consider both primary and secondary im-
316 RCA REVIEW
pedances. If we call the first of these Rpl and the second R9,,, the gain
of the first tube will be proportional to 1/2 \/Rp1 R92 . The mean -squared
noise voltage at the grid of the second tube will be proportional to
R,10/2 + R the primary load with optimum coupling halves
the value of R9, for noise-calculating purposes. Thus, the total effective
equivalent noise resistance referred to the grid of the first tube is, in
this case,
R9,,
+ Re`'-
2
Reg = Real + (14)
(GAIN) 2
In both Equations (13) and (14) the gain has been assumed to be the
grid -to -grid gain. It will be found that for given first and second
(a) SINGLE -TUNED CIRCUIT
1 po
o L
R2 TC2
( b) EQUIVALENT CIRCUIT
C2 = Cp1 + Cg2
Rp
R92
Ra= R
Pi 92
Fig. 7- Circuit and its equivalent for coupling first tube to second tube.
tubes, there is little to choose between the single- circuit and the
coupled- circuit cases represented by Equations (13) and (14) as
regards ultimate signal-to -noise ratio. This is in marked distinction
to the wideband analysis to follow.
B. Wideband operation: bandwidth a major consideration now -It
becomes necessary to consider the receiver in which the bandwidth of
the coupling circuit between first and second tubes is an important, if
not paramount, consideration. When the first tube is a converter or
mixer, for instance, the bandwidth of the coupling circuit is almost
always of primary concern since it is the first i -f circuit. When the
coupling circuit is a single -tuned circuit, as in Figure 7, the effective
bandwidth between points 1 db down on the curve is
1
Bandwidth pf' = (for 1 db down) (15)
47rC2R2
When the value of R2 from (15) is substituted in (13) and the band-
width is pf, the total effective noise resistance becomes*
+ 4irC2pf Reg2
Reg - Regi +
1
(16a)
4irC2p f (GAIN) 2
However, the gain is also expressible as gm1R2 where g,,,1 is the trans -
conductance (or conversion transconductance) of the first tube. Thus
(16a) can also be written (using again the value of R2 from (15) )
4irC. p f'
ReQ = Reg1 + (1 + 47rC2 A f Req2) (16b)
(gm1) 2
It is seen clearly that, for the second tube as well as for the first, and
when the bandwidth is a major consideration, the performance is
Fig. 8- Double -tuned circuit for coupling first tube to second tube.
specified by the product of a capacitance and the equivalent noise
resistance, or C2 Req2.
For wideband applications it is highly advantageous to separate
the two capacitances in an interstage coupler as is done by use of the
pair of coupled circuits of Figure 8 or its equivalent T or Tr network.
The fluctuation noise analysis in this instance is complicated for the
first time by the possibility of a triangular instead of a rectangular
noise spectrum. In a pair of coupled circuits (see Figure 8) adjusted
for flat -top transfer impedance (i.e., second derivative of gain vs. fre-
quency deviation set equal to zero), the impedance looking in at one
pair of terminals may be decidedly double- humped if the circuits are
of unequal Q. Thus the mean -squared thermal noise across such a
pair of terminals must be obtained by integrating the resistive com-
ponent of impedance over the pass band. Because the complete discus-
sion is unimportant at this point, the derivation will be given in
Appendix A and only the results discussed here. It is found that, by
having a low -Q secondary circuit and a high -Q primary circuit, a
* It is assumed that the total thermal noise is 4kT R2 pf. Actually if
f f = pf', due to dropping off of the tuned impedance near the edge of the
pass band, this would be about 8 per cent too high, a negligible correction.
318 RCA REVIEW
substantial reduction in thermal noise is made possible with no sacrifice

in gain or circuit bandwidth.
The effective bandwidth of a pair of coupled circuits (Figure 8)
taken between points 1 db down on each side and adjusted for flat
response is
1 1
Bandwidth pf' = - 1
47r Cp1 Rpl C R,_,
+ (17)
47r `K 1 Q_
where is the angular frequency at the center of the band and Q1 and
Q_, are primary and secondary Q's, respectively. As shown in the
appendix, the equivalent noise resistance at the grid of the first tube
is then
47rp f'Cp1
Re., =R + .,
X
gnil
Ql Qi
1 + 0.158 )'27rtfFCYRe q_
Q2
Q2
(18)
2
1-F
Q2
where is the transconductance or conversion transconductance of

g,,,1
the first tube. It is easily shown that (18) is lowest when Q1
(see appendix) and this choice should always be made whenever pos-
Q_
sible. The part of the noise due to thermal agitation in the coupling
network is reduced considerably by the choice. With this condition
(i.e., Q1 Q0)
47rp f'Cp1
Rev= RBQ1 + (0.15 + 27rof'Cp2Reg0) (19)
8miz
Again, as in the previous cases, it is evident that the noise of the second
tube is determined by the product C,2 Reg2 which should be as small
as possible. It is of considerable importance that the primary capaci-
tance, Cpl, (which is determined largely by the output capacitance of
the first tube) should be as small as possible. If the condition Q1
is fulfilled, it will seldom be found that the thermal noise of the inter-

stage circuit is appreciable compared with the noise produced by the
SIGNAL -TO -NOISE RATIO OF t -H -F RECEIVERS
T 319
second tube. Thus, for most practical purposes, (19) may be replaced
by
Reg_
Reg = Rr + (19a)
(GAIN)
provided coupled circuits are used as indicated. Equation (19a) is
obtained from (19) by neglecting the factor, 0.15, substituting IZ
for its equivalent, and finally letting g,,,1 = GAIN.
VIII -PRACTICAL APPLICATION TO U -H -F RECEIVERS
Examples of the application of the foregoing analysis are readily

made to ultra- high- frequency receivers for operation above 300 mega-
cycles. In the following applications, it must be emphasized that feed-
back effects have been neglected and that, to some extent, this neglect
may lead to discrepancies between measured results and those calcu-
lated herein. Furthermore, since the purpose of this discussion is
primarily illustrative, the tube data to be used will be only approx-
imate. Exact tube data at frequencies above 300 megacycles or so are
not available in any event.
The triode mixer, used in the converter stage of the receiver, will
form an important part of the discussion and it may be well to offer
a word of explanation. A triode in the converter stage, followed by a
low-noise i -f system, seems to offer the greatest promise for a good
overall signal -to-noise ratio with tubes now commercially available. In
the ordinary triode mixer, signal- frequency feedback from the i -f
circuit (which looks like a capacitance at the signal frequency) through
the grid -plate capacitance may be very important in reducing the input
resistance. However, in receivers designed to cover a very limited
tuning range, a simple expedient may be used to greatly reduce feed-
back. If the i-f output circuit is designed so as to present a very low
impedance to signal frequency, the signal- frequency voltage on the
plate may be reduced so that feedback becomes negligible. As a simple
example, consider the primary winding of the i -f transformer. At
signal frequency, this winding is above its natural resonance, and
behaves as a capacitance. If we add, in series with it, a small induc-
tance (a loop of wire or a length of lead) and adjust the latter to series
resonance with the effective distributed capacitance of the i -f coil, we
obtain a series resonant output circuit at signal frequency without
affecting, in any way, the i-f circuit. As a minor point in connection
with the triode mixer, the input capacitance and output capacitance
are each increased by the amount of the grid -plate capacitance.
320 RCA REVIEW
The equivalent noise resistances which will be used are taken from
formulas given in another paper.''- The conversion transconductance
is assumed to be adjusted to optimum value. The electronic input
conductance figures are based on available data with fixed voltages
applied and, in the case of the converter stage, are averaged over the
oscillator cycle. The capacitance figures which will be used represent
reasonable approximations.
The data will be presented in the form of a table. In the table, the
electronic input conductance is expressed in micromhos per (mega-
cycle) 2. Thus, to obtain the input conductance at any one frequency,
the constant given must be multiplied by the square of the frequency
in megacycles. In somewhat similar fashion, the non -electronic input
loss is expressed in micromhos per megacycle. The figure given repre-
sents only a rough guess, but since this part of the loss is small in
comparison with the electronic loss, its exact value is unimportant.
Circuit losses, if any, will be assumed to be included in the non-
electronic conductance figure. The Type 955 is assumed to be operated
at a plate voltage of 180, the 954 at a screen voltage of 100. The
column giving the peak oscillator voltage required is of importance in
estimating the power required from the local oscillator. The grid bias
of the mixer tube is assumed equal to the peak oscillator voltage in
all cases.
TABLE I
Equiv. Trans- Elec- Non- Approx.

Noise conduct. tronic Elec- Peak
Resist. gm or g. Per tronic gs2 Oscil-
First Tube and Method R qi. ge, (Mc),2 Per Mc., lator
of Operation Ohms mhos mhos mhos Volts
955 Mixer-At Oscillator 5,200 700 0.0030 0.2 5
Fundamental
955 Mixer -At Oscillator 12,000 350 0.0015 0.2 12
2nd Harmonic
3rd Harmonic
Fundamental
954 Amplifier 6,000 1,400 0.0050 0.2 -
Each example to be computed will be considered at three signal
frequencies, 300, 500, and 1,000 megacycles. The intermediate fre-
12 E. W. Herold, "The Operation of Frequency Converters and Mixers
for Superheterodyne Reception ", submitted for publication to Proe. I.R.E.
SIGNAL-TO -NOISE RATIO OF U -H -F RECEIVERS 321
quency (i-f) is chosen as 30 megacycles and the overall bandwidth of

the radio portion of the receiver is 7 megacycles, in order to minimize
tuning difficulties and to permit wideband operation, as for television.
However, when the relative sensitivities (signal equal to noise) are
computed, they will be referred to an arbitrary bandwidth of 10 kilo-
cycles. The actual sensitivity for other bandwidths following the
second detector, is readily computed by multiplying by the square-root
of the bandwidth ratio, in the usual way, with a negligible error.
A. The i -f tube and circuit -It is observed from Equation (19)
that the best tube to select for the i -f amplifier is one with the lowest
product of Cpl and Re42. It is easily determined that the type 6AC7/1852
is the best of the commercially available tubes in this respect, since it
-
has a value for Req2 780 ohms and an effective Cp,, (including circuit
and leads) which can be made as low as 20 micromicrofarads. This
tube will, therefore, be chosen. In each example, the tube working into
the i -f system will be an "Acorn" type and a total effective output
capacitance of 5 micromicrofarads (Cpl) will be assumed. This value
is high enough to include circuit and leads. There is no significant
difference between the 954 and 955 types in output capacitance because,
with the latter, it is necessary to add the grid -to -plate capacitance as
well. Because the i -f system is common to each example, we may com-
pute its contribution to the noise immediately. The output plate
impedance of a 955 mixer or converter is easily shown to be in excess
of 20,000 ohms, while the 6AC7/1852 has an input impedance of around
5,000 ohms. Thus, if we use a pair of coupled circuits as the first i -f
transformer, Q1 > 19 and the secondary loaded by the 6AC7 only will
have Q2' 19.f These Q's are so much higher than required to give
our desired 7- megacycle circuit bandwidth that a damping resistance
must be added.
Following the low -noise design exemplified by the discussion fol-
lowing Equation (18) we will add all the damping to the secondary.
For simplicity, let us assume Q1 = 19 in every case.$ Then Equation
(17) gives
-+-=2-=
1
Q,
1
Q2
La,
2
7
30
= 0.47
If Q1 = 19 we see that Q2 must be 2.4 and that we must add a shunt

resistance of 730 ohms across the input to the 6AC7. Under these
t Obviously the i-f transformer losses can be neglected.

$This implies adding a 20,000 -ohm shunt resistance across the primary
of the i -f transformer when a pentode mixer is used.
322 RCA REVIEW
circumstances, Q1 /Q2 = 7.9 and the transfer impedance is 1,600 ohms,

approximately. From Equation (18) we find the effective equivalent
noise resistance at the grid of the converter stage is
523 X 10-6
Reg = R,.,,1 +
got
Using the values from Table I for Regi and ge, we find Reg for each of
the converter stage possibilities.
B. Converter as first stage -It is first necessary to determine
whether the input circuit bandwidth is sufficiently wide. The narrowest
bandwidth of any example considered, will be obtained with the lowest
frequency, 300 megacycles, and the highest value of input resistance,
obtained with the 955 mixer operating at the third harmonic of the
applied oscillator frequency. Using Equation (4) and letting gaR1 =1
(the lowest value which should be considered in a low -noise design),
the circuit bandwidth is at least
1
Pf' _ - - -- __
rCR1
----
-7rX5X10--1'X
1
10i 6
= 9.6 megacycles
0.001 X (300) 2 + 0.2 X 300

Thus, it is found that the r -f bandwidth need not be increased in any
of the examples and the receiver may be designed for best signal -to-
noise ratio. Section IV of this paper is then applicable.
The calculations for each example are straightforward and only one
will be carried out in detail. The others will be tabulated later in this
section of the paper. We see from Equation (9) that we need only
Reg, R1, and ge. Let us choose for the detailed example the 500 -mega-
cycle receiver with a 955 mixer operating at oscillator fundamental.
Then
523 X 10-6
Reg = Reg1 -}- = 5200 -+ 1070 = 6270 ohms
got
Also
1 106
R1 = _ = 1180 ohms
ge gn 0.003 X (500)'= + 0.2 X 500
Finally
gel?' _ 0.003 X (500) 2 X 10 -6 X 1180 = 0.88
Using Equation (9), we find = wetz X 5.1
e0
For Ra = 75 and Pf = 10 kc, \/et2 = 0.11
microvolts so that the mini-
mum open- circuit antenna signal to equal the noise is ea =0.56
microvolts. If a signal generator is used which has been cllibrated
SIGNAL -TO-NOISE RATIO OF' U -H -F RECEIVERS 323
for the matched impedance condition, its calibration should indicate
just half of the above value.
It is interesting to compare the sensitivity just computed with that
which would obtain if the impedance- matched condition had been used.
Using Equation (6), we find
impedance- matched ea
min
= 0.57 microvolts
This represents a negligible reduction in signal -to-noise ratio. The
result should have been expected from inspection of the curves of
Figure 6 which are nearly coincident at the value of Ri /Req appropriate
to this example.
C. Amplifier asfirst stage -Let us assume a 954 amplifier followed
by a converter stage with the lowest noise equivalent (the 955 operated
at oscillator fundamental). The calculations will be made neglecting
the induced grid noise in the converter and assuming an idealized
condition with no feedback in the r -f stage. In this example, the inter -
stage circuit impedance will be given by half of the geometric mean
of the 954 output impedance and the converter input impedance. Recent
data on the 954 indicate an output conductance due to lead losses etc.
roughly equal to
954 output g = 0.0006 X (megacycle) 2 (in micromhos)
For this relation, the 300 -, 500 -, and 1,000 -megacycle interstage trans-
fer impedances are then 3,700, 1,400, and 360 ohms, respectively. With
,
g =1,400 micromhos, the grid -to -grid gains are then 5.2, 2.0, and 0.5,
respectively. The value of Reg2 will be taken as 6,270 ohms (to include
i -f noise) while ReQi (Table I) is 6,000 ohms. We find from Equation
(14)
Rg2
+ Reg=
2
6,300 ohms for 300 megacycles
= Reg, 6,700 ohms for 500 megacycles
Re,, -}
= 32,000 ohms for 1,000 megacycles
(GAIN) 2
Using the conductance data from Table I, we find
2,000 ohms for 300 megacycles
R1 = 740 ohms for 500 megacycles
190 ohms for 1,000 megacycles
The minimum open- circuit antenna signal is then readily computed.
For Equation (9) (the condition for best signal -to-noise ratio), the
relative open -circuit antenna signals (Ra = 75, f = 10 kilocycles) are
0.46 microvolts at 300 megacycles

ea = 0.70 microvolts at 500 megacycles
"" 2.9 microvolts at 1,000 megacycles
324 RCA PEVIEW
It is evident from these figures that a 954 r -f amplifier stage is

inferior to the 955 converter stage, from a signal -to-noise point of view,
even at 300 megacycles. This was already in evidence in Table I since
both Req1 and ge were lower for the 955 mixer than for the 954 amplifier.
TABLE II
Minimum Open- Circuit Antenna

Signals
First Tube 300 Mc 500 Mc 1,000 Mc
955 Mixer -At oscillator fundamental O.39 V 0.56V 1.02V
955 Mixer -At oscillator 2nd harmonic 0.45 0.66 1.20
955 Mixer -At oscillator 3rd harmonic 0.47 0.69 1.23
954 Mixer -At oscillator fundamental 0.75 1.15 2.22
954 Amplifier -Followed by 955 0.46 0.70 2.9
Fig. 9-Double-tuned circuit with constant-current drive.
D. Tabulated results for all examples -It is assumed that R0 = 75

ohms, P f = 10 kilocycles
and the design for best signal -to -noise is used
(Equation (9) ).
One of the more interesting aspects of the table is the fact that the k
955 used as a mixer at 2nd or 3rd harmonic of the applied oscillator
is less than 2 db poorer in signal -to -noise ratio than it is with the
oscillator fundamental. It is, of course, necessary to apply a consid-
erably larger oscillator voltage when harmonic conversion is desired.
APPENDIX A
Calculation of Noise in Wide Band Coupled Circuit Fed by

Constant-Current Source
Because the behavior of coupled circuits is reasonably well known,
no attempt to derive the basic relations for transfer impedance, band-
width, etc. will be made here. The relations which will be used repre-
sent the usual approximations which hold exactly only for high Q's
(i.e., bandwidth small compared to center frequency), but which are
SIGNAL-TO -NOISE RATIO OF' U-H -F RECEIVERS 325
qualitatively correct even for a low-Q circuit. Referring to the figure,

if Q1 = wCpl Rpl, the primary circuit Q, and Q2 = wCg2 Rg2, the sec-
ondary circuit Q, then the transfer impedance is
1
-
1
e1 V2 Q12
Q22
-IZtl=
i w0 CP1Cg2 1 1 2
+
Q1 Q2
1
(20)
2
1+
1 1 4
+
Q1 Q2
where Sw is the deviation from the center angular frequency, wo. This
equation assumes flat -top response, i.e., zero second derivative of
IZtI
at center frequency. The total bandwidth, z f', for one db down on
each side is then twice the deviation frequency for a one db decrease
in IZtJ and is
27rpf' = pu, = -
wo
2 Q1
1
--- -- 1
Q2
(21)
Using (21) in (20) we get
0.707
1 + /
\Q
-Q1 2
1
(22)
(-
2
= Sto 4
wVCp1Cg2 Q1 1+4
1+ Qw
Q2
The real part of the impedance looking into the Cg2, Rg2 terminals is
Q1 Sw
1+2
1 Q2 Qw
RI part of Zsecondary = (23)
Qi 8w 4
L wCg2 1 -I-
1 + 4 6,w
It is interesting to note that, if Q1 Q2, a curve of Equation (23)
ilk
326 RCA REVIEW
is decidedly double-humped as shown qualitatively in Figure 10. Since

the thermal noise is proportional to this variable, the noise spectrum is
non -uniform and, in fact, the noise at center band is very small. This
behavior results in a very low average thermal noise over the pass
band when f = tif'. It should be remembered that Pf, the effective
pass band for noise purposes, is largely determined by the system fol-
0o
lowing the final demodulation of an incoming signal whereas p f' _ --
2ir
is the circuit bandwidth. Since the maximum noise will occur when
of = of', the calculations will be made on this basis. Since, as will be
ham-- BANDWIDTH
I i
Sw /aw +
Fig. 10-The real part of the secondary impedance of a flat-top, double -
tuned circuit for frequency deviations around mid -band and with primary Q
much higher than secondary Q.
found, the thermal noise of the coupled circuits may be made very low,
the correction which might have been applied when L f
not have been of significance in any event.

p f' would
We are now in a position to compute the thermal -noise voltage at
the grid of the second tube, add to it the equivalent second -tube noise
voltage, and, if we choose, refer the total back to the grid of the pre-
ceding tube by dividing by the square of the gain, (g,,,1Zt) 2. The mean-
squared noise of thermal agitation at the grid of the second tube is
Ato
f
(0o
2
et22 circuit = 4kTR Rl part of Zs, d a)
tir A.
2
where wo is the mid -band angular frequency. This is equivalent to the

thermal noise of a resistance at the second-tube grid of value
r
SIGNAL -TO-NOISE RATIO OF U -H -F RFCEIVERS 327
1
Rt., Rl part of Z, d 0,
0O)
o - 2
zi
=2 Rl part of Z".,, d
0
where &a is the angular frequency deviation from the center-band point
and Po) is the angular bandwidth to points 1 db down. Using Equation
(23), we obtain
2 1 Ql
R12 2 1 -I- 2 x2
80,
Q2
Q1
dx where x =
Ao,Cg2 1 + 1+4x' Aw
Q.,
`qGl
1 -I- 0.158
0.96 Q2
(24)
Cg2 Ql
1 -F
Q2
varied from one extreme (Q1

Q1
Equation (24) shows that Rt., varies over a 6:1 range as Q1 /Q2 is
Q2) to the other (Q2 Q1) . When
Q2, the noise spectrum is non -uniform and, at mid -band fre-

quency, the thermal noise is very much reduced over that at the edge
of the band.* The transfer impedance (i.e., the gain) is, of course,
constant over the band by the a priori relations.
The total effective noise resistance at the second -tube grid is (R60,
+ R12) where ReQ2 is the equivalent noise resistance of the second tube.
The noise may be referred back to the first-tube grid by using (21),
(22), and (24) to get
Reg, + R12
Reg = Rol-}-
(gm 1Ztl)Z
* It should be emphasized that the thermal noise -reduction possibilities

of coupled circuits cannot be utilized in the antenna -to-grid circuit in the
same way.
328 RCA REVIEW
2pftCPl
=R1 +
gm12
096 1 -I-
Q1
1 -}- 0.158
Ql
- Q3
(1+.)2AwC2Rez 1
Q2 Q2 Q2
L (25)
2
(aZ l
1 +
Q2
The factor in the bracket of Equation (25) is determined by the
choice of Q1 /Q_, which is usually up to the designer. Curves of this
factor are plotted in Figure 11. They show that Q1 /Q2 should be high,
pv
W~
6
w
o
30 4
W W
Ln m
O
Zr 2
W Z
U
_O
< vi
0.01 0.1 10 IO 100
O
RATIO OF PRIMARY Q TO SECONDARY Q (o,/Q2)
Fig. 11 -The total fluctuation noise of a flat-top, double -tuned circuit of
bandwidth pc whose secondary is connected to a tube of equivalent- noise-
resistance, R.,,, and whose total secondary capacitance is Co.
i.e., the wide-band circuit damping should be concentrated in the

secondary. The overall noise improvement to be obtained by such a
choice is ordinarily considerably less than the theoretical maximum of
8 db because of the major role played by the first-tube equivalent noise
resistance, Reg, not to mention the noise of the second tube as expressed
in Reg. However, since there appear to be no major disadvantages to
the choice, it can be concluded that best design of the interstage
coupling circuit will occur with the damping on the secondary only.
APPENDIX B
Signal -to -Noise Ratio for Mixer and Converter Tubes with
Control Grid Not Adjacent to Cathode
In some mixer and converter tubes the input or signal grid is not
adjacent to the cathode and the preceding formulas must be modified
SIGNAL -TO -NOISE RATIO OF' U-H-F RECEIVERS 329
before they are applicable. Furthermore, the electronic conductance

of a tube of this kind is usually negative12 so that some consideration
must be given to this aspect. The analysis and measurements of
induced grid noise in such tubes gives the noise in terms of an equiv-
alent saturated diode current, Id.5 Thus the mean -squared grid noise
current is
4,2= 2eIdpf
and this noise current cannot be expressed in a simple manner in terms
of ge, the electronic input conductance, as when the control grid is
adjacent to the cathode. However, at frequencies for which the transit
angles are not too great, Id increases with the square of the frequency
just as does the magnitude of ge.
The formulas developed in the body of the paper assumed the
induced noise current to be 4,2= (5 X 4k TR ge p f) so that, for the
present case, we find that the quantity 4ge R1 of the formulas becomes
22
4k TR
n
At
ge R1 = (201,t - ge) Rl (26)
For the tubes we are now considering, R1 may be negative and it must
be appreciated that stability requirements demand that the reflected
antenna resistance be lower than R1. An even more stringent require-
ment is that imposed by bandwidth. This requirement is given by
Equation (4) and may be written
ga po)C - R1
1
(27)
where pu) is 2irpf' and a single-tuned circuit is assumed.

The value of ga for optimum signal -to -noise ratio, irrespective of
bandwidth, is given by Equation (7) . This optimum coupling condi-
tion is rewritten (using Equation (26) )
gat
1
= R12 +
1- geR1 -f- 20 Id R1
R1 Rev
1 g + 20 Id
(28)
R12 Rev
If the value of go from Equation (28) is larger than that from (27)
it is possible to obtain a signal -to-noise advantage by using (28).
Otherwise, condition (27) must be used to assure proper bandwidth.
5 C. J. Bakker, loc. cit.
12 E. W. Herold, loc. cit.
330 RCA REVIEW
Using (26) and the bandwidth condition (27), the minimum signal
which is equal to the noise is found from Equation (3) to be
ea min
= \/4h T,eRQpf
/ Ta
+
gs.2 + 20 Id + ( PwC ) Reg
(29)
TR 1
pwC---
R1
If, on the other hand, the bandwidth is not a limiting consideration,
we may use (26) and (28) in (3) to get the lowest possible noise and
find
e
min
-= V 4 TXRQpf
Ta Req Req Req
+2 - -+2 + gnR1 + 20 Id R1 (30)
Tie R1 R1 R1
When R1 is negative, as here contemplated, the sign of the inner square
root must nevertheless be taken as positive if the result is to be valid.
LIST OF SYMBOLS
C = Total capacitance of input circuit to first tube.
C 91 = Total capacitance of primary circuit of interstage trans-
former.
CD2 = Total capacitance of secondary circuit of interstage trans-
former.
C2 = Sum of Cpl and C. Total shunt capacitance effective across
a single -tuned circuit between first and second tubes.
e = Open-circuit signal voltage from antenna.
et2 = Open-circuit mean -squared noise voltage from an antenna
which can be considered at room temperature.
F = Factor by which double-tuned antenna transformer is better
than single -tuned as regards bandwidth.
of = Overall receiver bandwidth for noise purposes (usually de-
termined after final demodulation of signal) .
of' = Circuit bandwidth. This has a minimum value equal to L f.
Po= Overall angular circuit bandwidth; equal to Zapf'.
Su, = Angular- frequency deviation from center of a band -pass

circuit.
= Center frequency of band -pass circuit.
gQ = Reflected antenna conductance in secondary of input circuit.
ge = Electronic input conductance of first tube.
gSt
= Circuit and "cold" tube loading conductance of input to first
tube.
g,,,1 = Transconductance or conversion transconductance of first
tube.
k= Boltzmann's constant (1.37 X 10-23 joules per K).
m = Effective turns ratio or step-up of antenna -to-grid circuit.
Qi = Ratio of shunt resistance to reactance of primary circuit of
interstage transformer.
Q2= Ratio of shunt resistance to reactance of secondary circuit
of interstage transformer.
Ra = Radiation resistance of antenna.
R1 = Impedance of secondary side of antenna -to -grid circuit with
tube connected but with antenna disconnected.
Rp1 = Shunt resistance across primary of interstage transformer.

R Shunt resistance across secondary of interstage transformer.
R.)= Shunt resistance across single -tuned circuit as interstage
network.
ReQ = Total equivalent noise resistance at grid of first tube includ-
ing noise due to first tube, interstage circuit, and second
tube.
R8Q1 = Equivalent noise resistance of first tube at its grid.
Reg_, = Equivalent noise resistance of second tube at its grid.
T,= Temperature of a resistance equal to antenna radiation re-

sistance and having same mean -squared noise as antenna.
TR = Room, or ambient, temperature.
IZti = Magnitude of transfer impedance of interstage transformer.
THE ABSOLUTE SENSITIVITY OF
RADIO RECEIVERS
BY
D. O. NORTH
Rescmrch LaInur;ilurii.,. K('.\ \Ianuf,ietnringr, Goinlr,iio . lu .. Il,irii.uu. \. J.
Summary -The total random noise originating in a receiver has cus-

tomarily been described in terms of the equivalent noise voltage at the
receiver input terminals. A comparison of the signal-to -noise ratios of two
receivers working out of identical antennas is thereby facilitated, but only
so long as the coupling between antenna and receiver input is extremely
loose.
This paper describes a method for rating and measuring the noise in
complete receiving systems, antenna included. The proposed rating appears
particularly applicable to ultra-high -frequency services and, more generally,
to any service in which signal -to -noise ratio is made a prime consideration in
receiver design and operation.
A portion of the study deals with the properties of receiving antennas,
yielding as a by- product an alternative derivation of Nyquist's theorem con-
cerning thermal fluctuations in passive networks.
A formula for absolute sensitivity is developed, which shows how the
minimum usable signal-field strength is related to the operating wavelength,
the antenna directivity, the local noise -field strength, the receiver band-
width, and a number called the "noise factor", which is a basic measure of
the internal noise sources of the receiver.
THE "Standards on Radio Receivers" adopted and published by

the Institute of Radio Engineers in 1938, define the sensitivity
of a radio receiver as "that characteristic which determines the
minimum strength of signal input capable of causing a desired value
of signal output ".' So defined, the sensitivity is a measure of gain.
The same report' takes cognizance of the fact that random fluctuations
(noise) originating in the receiver set a limit to the useful sensitivity.
Leaving a precise definition to those who write standards, we shall
refer to this useful limit as the "absolute sensitivity ", indicate in an
elementary way its dependence upon certain properties of the receivers
and antenna, suggest methods for measuring it in the laboratory, and
consider its modification by noise induced in the antenna.
1 "Standards on Radio Receivers ", 1938, definition 1R36.

2Page 42, section 13.
3A more thorough consideration of this aspect is the purpose of a com-
panion paper by E. W. Herold, "An Analysis of the Signal-to-Noise Ratio
of Ultra- High- Frequency Receivers ", RCA REVIEW, Jan. 1942.
332
ABSOLUTE SENSITIVITY OF RADIO RECEIVERS 333
The conventional method of describing receiver noise in terms of an

equivalent- noise -side -band input, while possessing a limited utility, does
not ever permit an immediate positive judgment of signal -to -noise
ratio, nor even a comparison of two receivers, unless additional infor-
mation is given concerning the antennas and the means for coupling.
It will be seen, on the other hand, that a description of the absolute
sensitivity of a receiver, antenna included, provides a direct basis
for the judgment of a complete receiving system, and facilitates the
inter -comparison of receiving equipments, no matter how diverse their
characteristics.' A rating of this kind is particularly valuable in the
ultra- high- frequency field, where antennas are designed as an integral
part of the receiver, or, for that matter, in any service wherein maxi-
mum signal -to -noise ratio is an important factor in design and opera-
tion. It was during an attempt, two years ago, to find a sound basis
for the comparison of television receivers, that the considerations
reported below were crystallized.
MEASUREMENTS IN THE LABORATORY

Consider an ultra- high-frequency receiver constructed to work out
of a specific antenna. This implies a specific radiation resistance,
antenna reactance, and ohmic loss (such as might be present in a long
down -lead) . The important essentials of a laboratory simulation of the
real antenna, therefore, appear to be
-a
1 signal generator providing a known radio -frequency voltage
source, e, in series with
-a
2 resistor, Ra, whose resistance equals the radiation resistance
of the prescribed real antenna;
3- tunable reactance to simulate the Q of the real antenna; perhaps
4- additional losses to simulate unavoidable losses in the real
antenna, and possibly
-a
5 low -loss transformer to correct minor defects in the attempted
similitude.
In many cases, the real antenna is prescribed to have a low Q, to
permit reception over a wide range of frequencies. In this event, and
provided the antenna Q is small in comparison with the Q of the
receiver input circuit, item (3) is of minor importance. By the same
token, emphasis upon "low- loss" in item (5) becomes less severe.
Furthermore, item (5) will often be found incorporated in the receiver
proper, in which case the responsibility for making it loss-free lies
with the receiver designer.
' This same subject has recently received the capable attention of R. E.
Burgess, "Noise in Receiving Aerial Systems ", Proc. Phys. Soc., Vol. 53, p.
293, May, (1941) .
334 RCA REVIEW
Consider next methods for measuring the noise output. Present

I.R.E. standards prescribe measurement of noise after detection.'- This
method not only requires a known modulation for the signal generator,
but also makes the noise measurement a function of the frequency
response of the circuits which follow the detector. On the premise
that post- detection filters are, or should be made sufficiently flat and
broad to pass uniformly all frequencies offered by the pre-detection
part of the receiver, it is here proposed to measure noise prior to
detection. The natural point at which to insert a meter is just ahead
of the detector. But occasionally latitude is permissible, and the de-
mands made upon the meter and its position may thus be summarized.
1 -It must be preceded by all the filters which materially determine
the r -f and i -f selectivity, and must not, in itself, modify the selectivity.
2 -It must be preceded by all of the significant sources of random
fluctuations.
3 -Its response to both noise and signal (preferably, to any admix-
ture of these) must be known.
Item (1) is often too stringent, and may be relaxed in circum-
stances where it is known that the frequency spectrum of the noise
power presented to the meter input terminals conforms essentially to
the shape of the power-selectivity curve at that point. This point will
be elaborated later.
A schematic of such a laboratory arrangement is shown in Figure
1, which, in addition displays one of the sources of noise, namely,
thermal agitation in the resistor Ra which, at room temperature To, is
simulating radiation resistance. The quantity k appearing in the for-
mula,
et2 =4kToR pf (1)
is Boltzmann's constant. The t
quantity f we will consider related to
the overall selectivity curve in the usual way. That is, if fo be the
nominal operating frequency of the receiver, and if G(f) be the re-
sponse of a linear output meter to input signal e (f) , then
of =f G2 (f)
G2(10)
df (2)
By any of various methods which avoid overloading any portion of

the system, it is possible to find a signal e (fe) which produces (as
indicated by the meter) an output signal power equal in magnitude to
the output noise power. Were the receiver ideal in the sense that it
possessed no sources of noise save that originating in the dummy
2 10c. cit.
antenna, it would be found that

7(f0)= 4k To Ra /f (3)
A fundamental description of the noise generated in the laboratory by
a receiver equipped with a dummy antenna is, therefore, provided by
the dimensionless measure
e2(fo)
N- (4)
4k T0 Ra pf
The number N we shall refer to as the "noise- factor ". Expressed
as a number, it states the ratio of actual noise power to that developed
by an ideal receiver. Or, it may, of course, be given in decibels referred
to the ideal receiver as a zero level. That is, the lower limit of N is
unity, or zero decibels.
DUMMY ANTENNA
2 3 O 4
1 - SIMULATED RADIATION RESISTANCE, AND POWER SOURCES

2- SIMULATED ANTENNA REACTANCE, OHMIC LOSSES, AND,
IF NECESSARY, A MATCHING TRANSFORMER
3- RECEIVER, LESS DETECTOR AND POST - DETECTOR STAGES
4- OUTPUT METER
Fig. 1- Schematic of a receiving system with dummy antenna and output
meter for laboratory measurements.
This proposed method for quoting laboratory measurement of

receiver noise is particularly valuable for two reasons.
First, it is generally fairly insensitive to modifications in the selec-
tivity, as was hinted above. For, unless the frequency spectrum of the
noise power output differs materially from the power- selectivity curve,
e2 (f a) is proportional to a. The noise factor N thus provides a rea-
sonably just basis for the comparison of equipments possessing selec-
tivity curves widely different as to both shape and width.
Second, the quantity e2 /Ra is invariant in a transformation through
a loss -free transformer. Consequently, whenever there are no pre-
scribed ohmic losses in the antenna, and when the object of the
measurement is not simply to measure N, but rather, to adjust the
antenna coupling to an optimum which then produces the lowest pos-
336 RCA REVIEW
sible N, one need not be limited to a value of Ra precisely equal to the

radiation resistance of the prescribed antenna. One need only make
sure that the prescribed antenna Q is provided, and even this require-
ment loses importance when the Q is too low to affect materially the
overall selectivity. For all special, but highly significant measurements
of this kind, it becomes unnecessary to know either the voltage calibra-
tion of the signal generator, or the precise value of Ra. Since only
the quantity e2, Ra appears in the expression for the noise factor, it is
sufficient to know simply the power the combination will deliver to a
matched load. In view of this property, it is seen that the minimum
noise factor provides, also, a reasonably just basis for the comparison
of equipments working out of antennas possessing widely different
radiation resistances.
SOME PROPERTIES OF RECEIVING ANTENNAS
Although we have obtained a method for determining and rating

the noise of a receiver with dummy antenna, we are not yet in a
position to describe its absolute sensitivity in operation. We must first
learn two things : one, the connection between signal -field strength
and voltage e produced thereby at the open terminals of an antenna of
radiation resistance Ra; two, the rle played by noise picked up by
the antenna.
The rle played by antenna reactance is altogether silent and will
be ignored.
Consider two antennas as pictured in Figure 2. The reciprocity
law (which is only a dignified way of saying that transfer impedances
for passive networks are the same in both directions) assures us that
if i is the current fed to one antenna, and e the resulting open- circuit
voltage at the other, then
e -ki
and k is symmetrical in the two antennas. But if antenna (b) is trans-
mitting, its power output is
P =i2Rr,
And if antenna a) is receiving some of that power, its open -circuit
(
voltage will be proportional to the square -root of P. Therefore,

VP.=
It follows that k is proportional to '/Rb, and, being symmetrical in the
two antennas, must be proportional to V/Ra also. It is, in addition,
proportional to a space- attenuation factor and t- the product of the
directivities of the two antennas.
We are thus enabled to write the relation between signal field

strength and open-circuit antenna voltage. Let S2 represent the two
angular coordinates which define a direction in space. Let 4 represent
the azimuth in a normal plane, to define the direction of polarization
of a Poynting vector S([1,4)) which describes the signal field. Let
D2(0,0) be the function which describes the power- directivity receiv-
ing pattern (identical with the radiation pattern in unencumbered
Fig. 2 -The reciprocity law for two antennas with individual

radiation resistances R and R,,.
space) , and let it be normalized so that, averaged over all values of

S2 and r, D2 (0,0)= 1. It follows that
e2 = ARaD2(l,4) S(ll,4) (5)
The coefficient A is evaluated in the Appendix by an appeal to Nyquist's
theorem, for which an additional derivation is produced, proving (if,
indeed, proof be needed) its valid application to radiation resistance
as well as ohmic resistance. The result, Equation (15), is
A = X2/27r
where is the wavelength.
A2
Hence, =
Ra D2(5,4) S(52,)
e2 (6)
2x
When practical units are used, and A. is given in meters, the relation
between S (watts per square meter) and its associated field strength
E (volts per meter) is
S = E2/120ir (7)
X38 RCA REVIEW
The open -circuit voltage produced at the antenna terminals by a

signal field is then
l2 R.
e- _ D2 (S2,43)E2 (8)
2rr 1207r
The voltage produced by a noise field may be likewise expressed.

It is certain that if the antenna were exposed solely to the isotropic
equilibrium- radiation of an enclosure maintained at room temperature
To, the noise voltage would be
e'==4kToRapf
It is equally certain that this is not the case. The strength of local
noise fields has been, and will always continue to be, the subject of
much serious speculation and measurement.5 While it is known already
that both man -made noise and atmospherics decrease rapidly with A,
the "resolving power" of receivers is not yet sufficient to determine
what limit, if any, is approached. Furthermore, the development of
new communications equipment for shorter wavelengths will always,
of itself, implement the production of what must necessarily be
regarded, from certain points of view, as new noise fields.
Local noise fields may be crudely divided into three groups:
1- Strictly random fluctuations similar to those generated within
the receiver proper, and characterized by a proportionality between
noise power and bandwidth.
2 -Noise consisting of impulses, random enough, but occurring at
a mean rate too low in comparison with the bandwidth to be included
with (1) , and, therefore, characterized by a lack of proportionality
between noise power and bandwidth.
3 -All other unwanted fields.
It is beyond our present purposes to discuss the relative amounts
of service disturbance engendered by equal amounts of noise power in
each of these categories. This vastly complex problem is receiving
attention elsewhere.6 Concerning ourselves with a measure of the
5 For example, K. G. Jansky, "Minimum Noise Levels Obtained on Short -

Wave Radio Receiving Systems ", Proc. I.R.E., Vol. 25, p. 1517, December,
(1937).
R. K. Potter, "An Estimate of the Frequency Distribution of Atmos-
pheric Noise ", Proc. I.R.E., Vol. 20, p. 1512, September, (1932) .
G. Reber, "Cosmic Static ", Proc. I.R.E., Vol. 28, p. 68, February, (1940) .
6 C. M. Burrill, "Progress in the
Development of Instruments for Meas-
uring Radio Noise ", Proc. I.R.E., Vol. 29, p. 433, August, (1941) .
ABSOLUTE SENSITIVITY OF RADIO RECEIVERS :3:311
power alone, we note that the difference between two determinations

of the noise factor, conducted as described, one with a dummy antenna,
the other with the real antenna in the absence of a signal field, pro-
vides just such a measure. We may go even further and, simply for
convenience, express this measure in terms of an essentially fictitious
temperature Ta of local space, writing
e2 =4kTaRapf
for the net open- circuit voltage produced by the local noise field at
the antenna terminals, so that?
Ta
N (operating) = N + -1 (9)
To
The total equivalent noise voltage at the antenna terminals of a system

in operation may consequently be written
Ta
e2 =4kToRapf E N+- 1 (10)
To
Should it be ascertained that the local noise field consists entirely

of the first kind, T. will then be independent of bandwidth, but not
necessarily independent of operating frequency; and if the noise field
is not isotropic, T. will naturally be a function of antenna orientation
and directivity. To the extent that noise fields of the second and third
kinds are present, T. cannot even be presupposed independent of band-
width.
ABSOLUTE SENSITIVITY
If, by the term "absolute sensitivity ", we refer to the r -m -s signal

field strength which produces at the open antenna terminals a mean -
square voltage equal to the equivalent mean -square noise voltage,
then, from (8) and (10), that field strength is
2407r" 4k To Pf Ta
E2= N -1
2 D2 (S1,0) To
7 E. W. Herold, Reference 3, exhibits the general functional relationship

of N to the important sources of noise within the receiver, and to the circuit
arrangements dictated by the service for which it is designed.
340 RCA REVIEW
If E is conventionally expressed in microvolts per meter, in meters,

Pf in megacycles, and T set equal to 300 degrees Kelvin (room tem-
perature),
39L Ta
N+ -1 (12)
A3D2 ([1,0) 300
The expression shows clearly the distinct, prominent contributions
of receiver proper, antenna directivity, and local noise fields.
The economy of efforts to improve the laboratory noise factor is
seen to be bounded, in some cases sharply, by the existence of local
noise. The use of tuned antennas for broadcast reception is, in many
locations, a costly luxury. In some recent experiments, W. R. FerrisS
and the author produced a noise factor N = 3.2 in an attempt to deter-
mine the lower limit for television service at 100 megacycles. Such a
receiver would probably be unappreciated in metropolitan districts.
On the other hand, at higher frequencies, Ta appears to decline, as
stated before, while minimum N rises rapidly for a number of tube and
circuit reasons.3 There is surely much to be gained from improved
N's at very short wavelengths.
The economy of antenna structures needs similar consideration.
For half-wave dipoles in free space, D2 is not a function of wavelength.
For dipoles beamed by parabolas large in comparison with X, D2
(maximum) is proportional to A /A1=, where A is the area of the para-
bola's aperture. Even were the noise factor to remain fixed as one
moved towards shorter wavelengths, the absolute sensitivity would
certainly suffer if one continued to use dipoles and would, indeed, still
suffer, despite the switch to a parabola, unless the aperture of the
parabola were given an area comparable to the square of the initial
wavelength.
CONCLUSION
Because of the ease of interpretation, and the simplicity of meas-
urement method, it is hoped that the material presented above may be
considered as a basis for the adoption of a standard of "absolute
sensitivity" for radio receivers.
APPENDIX
Thermodynamic reasoning alone is sufficient to justify the appli-
cation of Nyquist's theorem3 concerning thermal agitation to any
310C. Cit.
Research Laboratories, RCA Manufacturing Company, Inc., Harrison,
8
New Jersey.
9 H. Nyquist, "Thermal Agitation of Electric Charge in Conductors ",
Phys. Rev., Vol. 32, p. 110, July, (1928) .
passive network. But it is, at least, interesting to develop the theorem

in detail for a microscopic dipole exposed to equilibrium radiation.'
Using electrostatic units, we know that the radiation resistance of
a dipole of length 1(1
X) in vacuum is11
8 (,rfl)2
Ra-- (13)
T 3 C3
where f is the operating frequency, C the velocity of light.

The energy density of equilibrium radiation at temperature T is12
h f3 df hf -1
P=
87r
C3
[kT_l]
wherein h is Planck's constant, k is Boltzmann's constant. But since
hf /kT 10 -4 for f = 1,000 megacycles and T = 300 degrees Kelvin, the
Rayleigh -Jeans approximation is sufficiently accurate. Thus,
87r kTf1
p = df
C3
But, being isotropic,
E2 Ez2 + Ey22 + E2 3E,22

P=
47r 47r 4;r
Hence, the mean -square field strength at the antenna in its susceptible
direction (z) is
e
32(Trf)2
Ez2 = kT df
3 C3
Therefore, the mean -square open-circuit voltage at the antenna ter-

minals is, with the help of (13),
8(irfl)2
e =E2/2 =4
x
kT df = 4kTRadf (14)
To evaluate the coefficient A in (5), we submerge any antenna in

equilibrium radiation and demand that the open- circuit voltage appear -
1 A more general derivation was given by Burgess, Reference 4.
11 For example, Hund, "High- Frequency Measurements", p. 399, (Mc-
Graw-Hill, 1933).
12 For example, Ruark and Urey, "Atoms, Molecules, and Quanta ", p.
58, Eq. 10, (McGraw -Hill, 1930) .
342 RCA REVIEW
ing at its terminals obey Nyquist's theorem. Since D2 (52,4) has been
defined to have an average value of unity, and since (in e.s.u.)
C E-
S=
47r
it follows that
CE2 2rA
4kTRadf -e=ARa=ARa
S = 4kTRadf
47r Jl2
whence
A2
A= (15)
27r
This result enables one to state a perfectly general relationship between

the "effective height" of an antenna, its radiation resistance, and its
directivity. The effective height h (1 x) is defined
h--E e
Hence, from (8), we have, in practical units,13
,\' R
h"(52,4)) = D`s
(S2,0) (16)
27r 1207r
We further find, in agreement with Burgess,4 that the mean -square

effective height averaged over all orientations with respect to a fixed
signal vector is then
A2 Ra
h2 = (17)
27r 1207r
13 The
quantity, 1207r, carries the dimension, ohms. It is, in fact, the
characteristic impedance of a strip x units broad of a transmission line con-
sisting of two parallel infinite planes x units apart, supporting plane -wave
transmission.
4
loc. cit.
Finally, with reference to (6), it is to be noticed that the power

an antenna can extract from a signal Poynting vector is proportional to
e2 A2
D2 (f2,4) S
Ra 27r
The radiation pattern of dipoles shorter than a half -wavelength is

sensibly independent of the length; this leads to the curious observa-
tion that such dipoles extract sensibly equal amounts of power, no
matter how short they are. In the language of physical optics this
would be known as a diffraction phenomenon. In the language of
atomic physics, we can only conclude that the capture-cross -section of
short dipoles for passing Poynting vectors is of the order of V.
AN OMNIDIRECTIONAL RADIO -RANGE SYSTEM
BY
DAVID G. C. LUCK
Research Laboratories, RCA Manufacturing- Company, Inc., Camden. N. J.
PART II - EXPERIMENTAL APPARATUS

Summary- Experimental omnidirectional ranges have been developed
and tested in flight at frequencies of 6425 kilocycles per second and 125
megacycles per second. In each case, a radiating system consisting of five
vertical antennas and a metallic ground mat was used. Each transmitter
was of a normal radiotelephone type, supplemented by a pair of balanced
modulators, an impulse keyer, and a set of modulation controls. Full moni-
toring of the effect of all transmitter adjustments was provided. Essentially
normal aircraft receivers and antennas were employed. Both cathode-ray
azimuth indicators and pointer -type deviation from course indicators were
provided.
1- INTRODUCTION
THE previous section of this paper' compared the various methods
available for radio guidance and showed the place in this art
occupied by devices which indicate directly, continuously, and
automatically the direction of mobile receiving craft from special fixed
transmitters. We have called such devices "omnidirectional radio
ranges." The previous part of the paper also gave a detailed explana-
tion of the principles of operation of one form of omnidirectional
range system and an analysis of the effects of various possible sources
of instrumental error. In what follows, familiarity with this explana-
tion is assumed in order to avoid repetition. The purpose of the present
part of the paper is to describe apparatus developed in the course of
experiments on this type of omnidirectional range. The results ob-
tained in these experiments and some methods of using the instrument
for aircraft guidance will be described in a third part, to appear later.
Experiments on an omnidirectional range system were begun at
Camden early in 1936, a frequency of approximately 61/2 megacycles
per second being chosen for reasons of convenience. The first flight
test, in November 1936, gave the expected automatic bearing indication
1 "An Omnidirectional Radio -Range System. Part I-
Principles of
Operation" by David G. C. Luck, RCA REVIEW, Vol. VI, No. 1, p. 55, July
1941. Hereafter referred to as I.
344
OMNIDIRECTIONAL RADIO -RANGE SYSTEM 345
in all directions, but the accuracy attained was not of a practically

useful order. Continuing work on details of apparatus and operation
resulted, by the Spring of 1937, in usefully accurate and stable omni-
directional bearing indication. Field tests were continued through
the summer of that year to complete the work.
In view of the strong general trend of aviation radio toward ultra-
high frequencies and of the success of the early work, it was decided
to undertake the development of an omnidirectional range to operate
at 125 megacycles per second and to have instrumental errors not
Fig. 1- 6425 -kc omnidirectional range.
exceeding three degrees. Work was begun early in 1938 and equipment
was built and installed by the latter part of that year. As there was
at that time considerable interest in the project, an attempt was made
to get the new system into operation early in 1939. As a result of the
accuracy required and the problems of ultra- high-frequency technique
encountered, however, quantitatively acceptable flight results were not
attained until the late summer of that year. Flight testing continued
until early in 1940 and the apparatus was kept in operative condition
until dismantled upon the closing of Central Airport at Camden, New
Jersey, at the end of July 1940. Testing being then practically com-
pleted, the system has not been set up again.
1
'
s
346 RCA REVIEW
2- DESCRIPTION OF APPARATUS
A- Medium Frequency (6425 Kilocycles Per Second)
a-- Antenna System
Figure 1 is an aerial photograph in which the radiating system
used at the medium frequency is shown in relation to its surroundings.
The two power lines in the background and the swamp in the fore-
ground, in conjunction with the general slope of the ground toward
the swamp, made the location far from ideal for accurate directional
Fig. 2- 6425 -kc range antenna system.

radio work. The northeast corner of Central Airport, Camden was
just off the lower left corner of the picture.
The wooden -lattice tower supported five 30 -foot vertical wires
accurately located at the corners and center of a 12 -foot square, with
diagonals accurately aligned in true north -south and east -west direc-
tions. A counterpoise, erected in a level plane tangent to the highest
part of the sloping ground, comprised a 24 -foot square of expanded
copper sheet, fringed with 96 equally spaced radial wires extending
75 feet from the center of the antenna system. The circular area cov-
ered by the counterpoise shows clearly in the picture.
The transmitter was located directly at the center of the base of
the antenna square, as Figure 2, taken before erection of the counter-
poise, shows. The central, broadcast antenna terminated directly at
OMNIDIRECTIONAL RADIO-RANGE SYSTEM 347
the top of the transmitter. Each corner antenna was connected to the
center of the square, at ground level, by a short radial length of coaxial
line with the outer conductor grounded immediately under the antenna,
but otherwise insulated. Outer conductors were continuous along each
full diagonal of the antenna square ; the inner conductors were split
at the center of the square and there fed by short balanced lines from
the transmitter output terminals to ground level. These highly sym-
metrical directive antenna circuits were ungrounded, and they appar-
ently maintained antenna balance to ground quite well. Antenna height
Fig. 3- 6425 -kc experimental range transmitter.

was about 1/5 wavelegth and spacing 1/10 wavelength at the operating
frequency. b-
Transmitter
A very simple transmitter was used, shown by Figure 3. On the
bottom shelf will be seen power -supply and impulse- keying equipment.
On the shelf above, speech modulating equipment was installed later.
The second shelf from the top carried a crystal oscillator and two driver
amplifiers, feeding radio -frequency excitation upward to the 150 -watt,
broadcast -channel pentode amplifier on the top shelf and downward
to the directive-channel balanced modulators on the middle shelf.
Parallel grid excitation, push -pull plate modulation with only alternat-
348 RCA REVIEW
ing plate voltage applied, and push -pull output were the operating
conditions of the balanced modulators.
Two -phase, 60 -cycle power with good waveform, for the balanced
modulator plates, was derived from the single -phase supply line by
shunting one plate -transformer primary with a capacitor and feeding
the combination through a variable series inductor. This inductor was
made from a small Variac by introducing an air gap into its core. The
whole combination was fed from a standard Variac giving amplitude
control. This scheme worked well, but the control of phase and ampli-
tude by the two Variacs was not independent and readjustment was
necessary whenever balanced -modulator tuning changes altered the
load.
Fig. 4-Station -announcing phonograph.

A novel item in the speech equipment was the simple film phono-
graph shown in Figure 4, which repeated the 3- second announcement
"Central Airport, Camden, New Jersey" at 15- second intervals. A
normal 16 -mm sound track projected beyond the edge of the large disc,
around which it was wrapped. It ran, at normal linear speed for this
type of film, past a fixed scanning line of light. Modulated light passing
through the sound track was reflected, through a geared sector mask
to cut out four of every five announcements, to a phototube in the
cylindrical shield. A high -pass filter, with a 60 -cycle rejection point,
was interposed between the speech amplifier and speech modulator to
minimize 60 -cycle hum modulation of the broadcast carrier signal.'
c-Monitoring
Four carefully compared radio-frequency ammeters, installed at the
bases of the four corner antennas, served both to show that antenna
2 I-Sect. 3, C, p. 76.
symmetry was maintaining equal currents in the two antennas of each

diagonal pair and to permit adjustment of north -south and east -west
directive -pattern amplitudes to equality.3 One of the two windings
of a dynamometer meter was connected across each balanced -modulator
plate- transformer primary, so that the desired condition of phase
quadrature between the two modulating voltages' resulted in a null
indication on this meter.
Phase of the reference keying was not controlled in the medium -
frequency set -up, but was merely initially set to a convenient value,
the particular value being of no importance in a single transmitter so
long as it remains constant between receiving- indicator zero checks.5
The desirability of being able to get such zero checks at any time was
not at first realized, so no provision was made for non-directional trans-
mission of calibrating modulation. Calibration was accomplished by
turning off the balanced modulator delivering the smaller signal at
the receiving point, thus causing the receiving indicator to read that
cardinal bearing which lay nearest the true receiver bearing. Provision
was made, however, for supplementing the keying impulses -normally
occurring once per modulation cycle -with three others to give four
equally- spaced impulses per cycle for use in making complete receiving
indicator adjustments.
Toroidal pickup windings were placed around the bases of the
antennas and connected directly to the plates of a cathode -ray oscillo-
graph through short unloaded lines, for monitoring radio-frequency
adjustments. By connecting diagonally opposite toroids together in
parallel, their electrostatic coupling to the antennas was cancelled.
Highly characteristic pattern shapes on the oscillograph indicated
phase balance of the two tubes of each balanced modulator and phase
quadrature between broadcast and directive antenna radio -frequency
currents.6 Radio -frequency adjustments were always made with the
regular transmitter tuning controls.
d-Indicators
Regular communication antennas on several aircraft were used for
reception, the receiver being an aircraft superheterodyne with crystal -
controlled oscillator. This receiver was modified by being given a very
flat automatic gain control, broadcast -type intermediate -frequency
transformers to pass the rapid signal changes of the reference keying,
and an audio -output transformer having a characteristic which was
3
I-Sect. 3, C, p. 65.
4 I-Sect. 3, C, p. 75.
5 I-Sect. 3, C, p. 74.
I-Sect. 3, D, p. 77.
350 RCA REVIEW
flat down to 60 cycles. Otherwise it was an entirely normal aircraft

receiver, which simply fed audio output from its headphone jack to
the indicators.
In the cathode -ray indicator7 two loosely -coupled circuits, tuned to
60 cycles per second, selected the pattern- rotation component from the
receiver audio output. Output from this filter was applied directly to
one set of deflecting plates of the electrostatically deflected cathode -ray
tube, and through a 90- degree phase shifter of adjustable phase and
amplitude characteristics to the other deflecting plates. Impulses were
selected from the receiver output by a simple high-pass filter and
applied to the grid of the cathode -ray tube, momentarily cutting off
the electron beam and darkening its trace on the screen of the tube.
Accelerating voltage was applied to the cathode -ray tube, from a
vibrator unit with vacuum -tube high -voltage rectifier, through the
normal brightness, focus, and pattern -centering controls.
Thus, the cathode -ray indicator contained only two tubes, the high-
voltage rectifier and the cathode -ray tube itself. The latter was mounted
in, and magnetically shielded by, a conical housing of thin Nicaloi.
The indication given consisted of a circle of green fluorescent light,
with a narrow dark gap moving around it to represent motion of the
receiving craft around the transmitter. This indication was read
against a circular engraved- celluloid scale pressed against the tube
face. A green gelatine light filter was placed behind the scale, to
improve contrast under adverse lighting conditions by suppressing
much of the reflection from the white tube face without markedly
dimming the green cathode -ray pattern. Setting of the scale zero was
accomplished by turning the scale in front of the tube, or in the rough
initial stage by turning the tube in its mounting. Accuracy of reading
at points other than zero was assured by adjusting vertical and hori-
zontal centering and deflection phase, with a special four- impulse -per-
cycle calibrating transmission, until the four dark marks then present
indicated correctly the four cardinal points; finally, relative amplitude
of vertical and horizontal deflections was adjusted by eye to give the
best -appearing circle.'
A deviation -from -course indicator was provided as a separate ac-
cessory unit.' Pattern- rotation sinusoid output, from the tuned filter
of the cathode -ray indicator, was applied to a universal phase shifter
and thence, in push -pull, to the suppressor grids of a pair of pentodes.
Positive impulses were applied to the normally cut -off control grids of
these tubes in parallel. A differential plate-current meter indicated
deviation of the impulse phase from the instant of equal suppressor-
7 I-Sect. 2, B, pp. 67-70.

s I-Sect. 3, F, p. 80.
9 I-Sect. 2, C, pp. 70-72.
grid biases, which was in turn determined by the phase-shifter setting

and by the bearing of the receiver from the transmitter.
An auxiliary arrangement for aural deviation indication was tried,
but was not entirely satisfactory. This employed an unsymmetrical
multivibrator to connect a pair of headphones alternately to each of the
differential impulse tubes mentioned above. The connection was made
for a relatively short time to one tube, and for a much longer time to
the other. On course, a steady 60 -cycle impulse tone was heard, while
off course to one side telegraphic dots of this tone were heard pre-
dominantly, and off course to the other side dashes were heard. Avoid-
ance of timbre changes in the on- course tone was difficult, and in any
case the tone was not very markedly audible and was hard to dis-
tinguish from ignition interference.
Fig. 5-125 -Mc Adcock antenna system.

B- Ultra-High Frequency (125.0 Megacycles Per Second)
a-Antenna
At the ultra-high frequency chosen for our later tests, with a wave-
length just under eight feet, the Adcock antenna system consisted of
five two -foot rods, set at corners and center of a seven -inch square, and
a fifteen -foot diameter copper ground disc, as shown in Figure 5. This
system was set up in the middle of a hangar roof and, to avoid ill
effects from imperfect bonding of the metal structure of the roof, was
surrounded by a fringe of expanded copper sheet to give an overall
ground -plane diameter of 45 feet. A vertically polarized antenna sys-
tem was chosen because it was felt that a set of vertical rods could
be more easily designed, built, adjusted, and fed than the corresponding
set of horizontal loops which would have been necessary to provide a
horizontally polarized signal.
Across the front of the hangar, about 60 feet from the antenna
system, there was a brick parapet over 1/2 wavelength high and about
352 RCA REVIEW
six wavelengths wide. To facilitate measurements on the roof, the

energy approaching this barrier was deflected upward by a large
inclined sheet of copper netting interposed between parapet and an-
tennas. Metal uprights supporting obstruction lights were made de-
mountable and taken down during operation to avoid re- radiation. A
brick chimney, bearing about fifteen degrees from the antennas, also
gave rise to standing waves on the roof, but no practical way to avoid
this disturbance of the field pattern was found.
All antennas were insulated from ground at their bases and fed,
through the ground plate, by the system of coaxial lines inside the
Fig. 6-125 -Mc antenna feeding system.

hangar, as shown in Figure 6. The short straight line -segments shown
immediately below the circular antenna base -plate were quarter -wave
impedance transformers. With each antenna length adjusted to reso-
nance, these transformers were designed to have the correct charac-
teristic impedances to match the antenna base resistance (only three
ohms for each directive antenna because of the proximity of the other
antennas) to the approximately 60 -ohm impedance actually charac-
teristic of commercial 7/8 -inch coaxial line. Curved line- sections seen
joining the bases of diagonally opposite transformers were half-wave
stabilizing and decoupling elements, operating as described below.
The five lines running out of the picture were the matched feeders
bringing power from the transmitter, which was about eight wave-
lengths distant.
Considering one directive antenna pair only, its feed system is

shown in section by Figure 7 (a) . The antennas AB and FG were
shorter than a quarter wavelength, by the small amount necessary to
produce resonance or zero reactance at the base. The conditions to be
fulfilled were that radio -frequency currents at the antenna bases B
and F should always be equal in amplitude and opposite in phase.10
Introduction of the quarter-wave line- sections BC and EF not only
inverted the low antenna base resistance appearing at points B and F
to a chosen higher value at the feeding points C and E, but also insured
the fulfillment of the above antenna -current relations if equal ind
opposite radio -frequency voltages were maintained at these feed points.
A
e

1
%
PLATE
,....,., e
... ../
GROUND
F 1
I / 1
, I / 4 FROM TRANS. ,
I E 1
I
ii
K\"
..
(a)
Fig. 7- Operation of ultra- high -frequency antenna feed.

By use of the well-known phase- inverting property of half -wave trans-
mission lines, equal and opposite voltages were constrained to occur at
points C and E by interconnecting them through the half -wave line
section CDE. In the normal case, when the feed lines from the trans-
mitter were well balanced and were supplying equal and opposite
voltages at C and E, the line CDE simply bore a standing wave and
transported no power. Should point C, for example, tend to be fed at
a higher voltage than point E, the half -wave line would carry power
from C to E to oppose this unbalance and maintain the required phase
and amplitude relations.
Midway between the antennas of a directive pair there was also a
non -directive antenna, as shown at HI of Figure 7 (b) . Current in this
antenna tended to induce equal and co-phasal currents in the adjacent
lo I-Sect. 3, D, p. 78.
354 RCA REVIEW
antennas AB and FG; this result was not in itself particularly harm-
ful, but was found quite annoying because it made the input impedance
of the non -directive antenna depend on the circuits feeding the direc-
tive antennas. Consider for the moment that the half -wave stabilizing
line CDE is cut at point D and the feed line to point C from the trans-
mitter is disconnected. Then the quarter -wave line CD, open at D,
appears as a short- circuit at C; thus, the impedance at C is zero inde-
Fig. 8 -U -H -F omnidirectional range transmitter.

pendent of any feed line which may be connected there. The quarter-
wave line BC, effectively short- circuited at C, appears as an open
circuit at the antenna base B, so that any voltage induced in the
quarter -wave antenna AB can cause no antenna current to flow to
ground. By symmetry, the voltages induced in antennas AB and FG
by current in HI are equal and in phase, so that the induced voltages
transferred to the hypothetical cut at point D by the lines BCD and
t
FED are identical and this assumed cut in the Mine can be mended
without altering anything. Thus, the half-wave line section acted to
short -circuit both point C and point E for any co-phasal voltages
appearing there, at the same time as it equalized any contra-phasal
voltages there applied.
b- Transmitter
Figure 8 shows the appearance of the 125 -megacycle range trans-
mitter. Its general arrangement is shown in the block diagram of
Figure 9, where heavy lines represent radio-frequency paths, double
lines modulation paths and light lines power supply. Above the dotted
line appears a normal 200 -watt, 125 -megacycle radiotelephone trans-
mitter. This employed a third- harmonic, low temperature -coefficient
120 v.
30 PH.
60 _gyp RUN
SP
WR. PHON MOO.
AMP. CAL
MIC
X TAL O C
F REO.
aurc'ER
_L
DRIVER P. A.
MULT.
B AL
MOD.
DRIVER
IMP.
GEN
BAL
MOD
PHASE PHASE PHASE AMP
CONT. CONT. CONT. CON T.
o
CAL.
Fig. 9 -Block diagram of omnidirectional range.
crystal controlling a pentode acting as oscillator and electron -coupled
frequency doubler, a full -wave doubler stage, a push -pull tripler, and
a neutralized push -pull buffer amplifier as its exciter system. This
exciter energized neutralized driver stages operating at final frequency
and a neutralized push -pull power -output stage was modulated up to
50 per cent by a class -A audio power amplifier. An RCA -804 was used
as the oscillator, four 845's as the modulator, two 806's in the power -
output stage, and pairs of RCA -834 tubes in all other stages. Grid -leak
bias was used on all radio -frequency stages, with just enough cathode -
resistor self-bias to prevent overloading in the absence of excitation.
For tuning, all 125 -megacycle circuits employed "trombone" variable
inductances, resonated by tube-plate and neutralizing capacitances.
Below the dotted line of Figure 9 are indicated the elements dis-
tinguishing the omnidirectional range from an ordinary transmitter:
356 RCA REVIEW
a pair of balanced modulators, an impulse keyer, and their associated

power controls. Figure 10 illustrates the circuits of a neutralized
amplifier and one of the pair of balanced modulators driven by it; the
other modulator was identical with the one shown. The sliding contacts
which permitted inductance variation may be seen to have been at
positions of relatively low current in the standing waves on the elec-
trically long inductance trombones. Only alternating voltage was
applied to the balanced modulator plates. Small, differentially variable
condensers were connected between modulator -tube plates and ground
to permit balancing of the modulator output. These were adjusted by
watching the pattern on a cathode -ray oscillograph tube with one pair
r- I
BUFFER AMPI I PLATE CCT
I COUPLED I
TO CAR PA
PLATE CIRCUIT'
CRO
I NS BAL. MOD.
TO
DAR
ANT
L _ 7
a FROM
NS MODULATION
l 4
PHASE CONTROL
+1250
Fig. 10-Balanced modulator.
of plates coupled to the modulator output, as shown, and the other
pair coupled to the carrier power -amplifier output. In general,
the oscillograph pattern was a patch of dim light bounded by two
crossed ellipses. Variation of the phase of the steady reference wave
from the carrier amplifier, by adjustment of the tuned circuit shown
in one of the deflecting circuits, ordinarily permitted either bounding
ellipse (but not both together) to be collapsed to a straight line.
Correct adjustment of the differential capacitor, giving equally phased
outputs from both tubes of the modulator, was the only condition that
permitted both ellipses to be closed at once.
Thermocouple radio -frequency voltmeters were connected across the
feed lines to the directive antennas and served as antenna -circuit tun-
ing indicators. Some care was required in installation to avoid spurious
readings of these indicators caused by standing waves on the feeders
at harmonic frequencies. The output required from the balanced mod-

ulators was surprisingly low. Directivity, both vertical and horizontal,
of spaced antenna pairs gave a geometrical power gain of 21/2 with
respect to the single carrier antenna ; thus, an unmodulated directive
signal of maximum strength equal to that of a 200-watt broadcast
carrier would require only 80 watts into one pair of directive antennas.
Since about 40 per cent modulation was found convenient, only 16 per
cent of the latter power was required, or a peak output of 12.8 watts
from each balanced modulator. Modulation being sinusoidal, average
output was one -half the peak output, and the average power require-
ment was shared equally between two tubes in each balanced modulator.
Thus, the average output power required per tube was only 3.2 watts,
., .s.
FROM IMP
SE L. ADJ
I
D06+ 1111-4-
r ROM
IMP
PM COW
TI TO
D.C.
SP.
AMP.
NEGATIVE '
KEYING IMPULSES TO
TRIPLER ORIO ANO
IMP. PHASE INDICATOR
Fig. 11- Impulse keyer circuit.

and the peak per tube 12.8 watts, to provide 40 per cent modulation
of the 200 -watt broadcast carrier, successively in all directions. Actu-
ally, sufficient output to modulate a two-kilowatt carrier was easily
obtained from the RCA-834 balanced modulators, without approaching
limiting ratings or encountering serious instability. At the time this
work was done, tube limitations at high frequencies prevented a two -
kilowatt carrier output from being obtained economically at 125 mega-
cycles per second and caused the adoption of the 200 -watt figure.
Directive modulation was set at the desired level, which was not in
any way critical, by choice of balanced -modulator output coupling.
Figure 11 shows the circuit of the impulse- keying equipment. Key-
ing phase was determined from the alternating component of the out-
put of the full -wave rectifier supplying plate power to the speech
amplifier. The rectifier output was connected to a series of circuits
tuned to 240 cycles per second and terminating in a peaking trans-
former, of the usual type employing an easily- saturated lamination in
one core leg, with push -pull output. Each side of the peaking- trans-
former output fed one grid of a twin triode, so that with both grids
358 RCA REVIEW
biased just to cut -off in the absence of input, there appeared across the
common plate -load inductance 480 equally- spaced voltage impulses of
double polarity per second, or eight per modulation cycle. Accurate
tuning of the 240 -cycle circuits was essential to assure uniform spac-
ing of these impulses. By biasing one triode completely off and adding
a modulation- frequency component to the bias of the other triode, it
was possible to select one of the eight impulses, which appeared alone
60 times per second, or once per modulation cycle, in the common triode
output. These impulses gave sharp control of the discharge of a
capacitor through a non -oscillating gas triode, which improved their
shape without affecting their timing. Positive impulses from the gas-
FROM MOD PA:

FROM SP. AMP
Ii
CAL MOO PHASE

R
--I
TO F W
BAL. MOD. TO NS
E MOD. AMPLITUDE
-41
N5 MOD
' BAL. MOD.
PHASE
FROM R. F. MONITOR
EW MOO.
0' GOti NS MOO

CAL. MOO
3; SUPPLY KEYING IMP
Fig. 12-Modulation control and monitoring circuits.
triode cathode circuit were amplified to give large negative impulses

for keying the grids of the tripler stage. These rendered this stage
inactive for periods of rather less than 100 microseconds, accurately
recurring once or eight times per modulation cycle at will.
Two -phase supply for the balanced modulators was easily obtained
in the case of the ultra- high- frequency transmitter, since this took its
power from a three -phase line. As shown in Figure 12, the east-west
balanced modulator was fed directly from one phase of the supply,
through an amplitude adjusting Variac. The north -south modulator
was connected between the variable tap of a Variac, across the east -
west supply phase, and the third supply line ; this Variac permitted
phase adjustment over a x-30 degree range with the desired quadrature
condition at midrange. The small vector diagram in the figure shows
the phase and amplitude relations involved. Calibration and reference -
keying phases were controlled in just the same way as the north-south
modulation phase." It will be noted that the transformer feeding cali-

brating signal to the speech modulator was provided with a series re-
sistor and shunt capacitor. These permitted initially bringing the
proper Variac setting to midrange, despite phase shift in the speech -
modulator circuits; they at the same time permitted setting the cali-
brating modulation at the same level as the normal directive modula-
tion. In the circuit supplying the impulse -selecting bias voltage to the
impulse generator, a resistance -capacitance network permitted initially
selecting for normal range operation, from the eight impulses per
modulation cycle provided by the peaking transformer, the one that
allowed the impulse -phasing Variac to run nearest mid -range. This
control system worked very smoothly in itself, but sudden line phase
changes, probably due to changing load distribution on the supply sys-
tem, were sometimes observed. All modulation switches were ganged,
as indicated in Figure 12, to permit changing over from normal direc-
tive running to non -directive calibrating transmission with a single
operation.
The simple film phonograph of Figure 4 was remodeled by removing
the sector disc, adding to the phonograph disc a cam to key the Camden
telegraphic identification signal KM, and adding a set of time switches.
It then automatically transmitted a telegraphic identification signal on
each minute, a telephonic identification on each intermediate half -
minute, and a brief calibrating signal on each quarter-hour. The motor
being synchronous, these signals were usable for checking clocks in
aircraft. Telegraphic transmission, either by hand or by cam, was ac-
complished by energizing a relay to change the twin -triode bias, in the
impulse generator of Figure 11, from the single impulse to the eight-
impulse-per -cycle condition.
It will be seen from the above figures and description that an omni-
directional range transmitter is rather simpler than the pair of separate
transmitters which make up a single present-day low-frequency "simul-
taneous" radio range. The omnidirectional range is, in effect, the voice
channel alone of a simultaneous range, plus a pair of low-powered
balance -modulator stages and an impulse keyer. The radio -frequency
goniometer and course -bending artificial lines of the course -type range
are not necessary with an omnidirectional range. It is, however, very
convenient to be able to simulate their action, to a limited extent, to
compensate for small differences between individual transmitters
and installations and for slight variations which may occur from time
to time in a given installation. This we did, without interrupting oper-
ation, by the simple set of modulation- frequency phase and amplitude
controls described above.
" I-Sect. 3, C, pp. 74-76.
360 RCA REVIEW
c- Monitoring
Instruments were provided to monitor separately the effect of each
modulation -phase or amplitude control; these instruments as well as
the control circuits are shown in Figure 12. For the north -south direc-
tive modulation and the calibrating modulation, simple center -zero
dynamometer meters served to indicate phase quadrature with the east -
west modulation, the desired operating condition, by the absence of
deflection which is characteristic of such meters when their two wind-
ings are fed with currents 90 degrees apart in phase. Differently sensed
deflections corresponded to phase differences more and less than 90
degrees. The meter circuit constants were chosen to make indications
independent of small circuit variations and consequently highly reliable.
NORTH FIELD NS
PICKUP r - - - - -1
MOO LEV.
FROM N5
BAL MOD
INPUT
EAST FIELD
PICKUP
FROM Ew
BAL MOD
INPUT
Fig. 13 -Field monitoring circuits.
Because of the large ratio of root -mean -square to average value of

the short, widely- spaced reference impulses, an attempt to indicate
their phase by a dynamometer meter resulted only in overheating the
meter without producing useful deflection sensitivity. However, a dif-
ferential electronic circuit, of the type shown associated with the im-
pulse control of Figure 12, gave good results. In proper operation, the
reference-modulation impulses occurred just as the east -west modulat-
ing voltage passed through zero; that is, just when neither rectifier of
the monitoring circuit passed any currnt from the sinusoidal supply.
Thus, both rectifiers then passed the same impulse current, had the
same average output and, therefore, produced no deflection of the dif-
ferential meter. With improper impulse phasing, one rectifier output or
the other predominated and the meter deflected in the corresponding
sense.
The use of cathode -ray oscillographs to monitor radio -frequency

phase equality of the outputs from the two tubes of each balanced
modulator has been described above. It is also necessary to monitor
the radio -frequency phase relations between directive and non -directive
radiated fields.12 Electronic wattmeter circuits located at and coupled
to the antenna bases gave very sensitive and convenient indications-
Fig. 14-Monitor unit.

of the zero-center sensed type -of departures from the desired condi-
tion. However, this system was found to be too dependent on indicator
circuit conditions to have the degree of reliability necessary for satis-
factory monitoring.
Antenna-current phase was finally monitored by the system shown
in Figure 13. Small tuned -loop pickups with diode rectifiers were
placed due north and east of the antenna system and distant by about
12 I-Sect. 3, D, p. 77.
362 RCA REVIEW
one wave -length, so that each was in the null of one directive signal
and was subject only to the broadcast carrier and the other directive
signal. Carefully grounded and shielded cables carried the low -fre-
quency output from these units, visible at the edge of the ground disc
in Figure 5, to the transmitter. There, the direct-current components
from the two pickups were added, to give an indication of the non -
directionally radiated carrier level, and the alternating components
were fed (across differentially adjustable resistors to equalize the
pickup unit sensitivities) to separate, modulation- frequency narrow -
band -pass filters. The filter outputs were separately rectified and
metered, by microammeters which indicated the two directive modula-
Fig. 15.-U-H-F receiving antenna on airplane.
tion levels, and by dynamometer meters which checked field-modulation

polarity with respect to modulating input and so guarded against gross
radio-frequency phasing errors.
Modulation depth being greatest when carrier and suppressed -
carrier modulated fields are in phase, phasing adjustments were made
by means of the transmitter tuning controls to maximize the modula-
tion -level indications while maintaining the correct modulation polarity.
Close spacing of transmitting antennas and monitoring pickups caused
the fields at the pickups to differ from those at great distances and
consequently resulted in the radio -frequency phasing being incorrect by
about 10 degrees. This effect, being the same for both directive fields,
caused no direct bearing errors,13 whereas greater pickup spacing re-
sulted in excessive disturbance by nearby obstructions. Modulation-
level meter sensitivities were equalized, during calibrating transmis-
sion when both pickups were known to receive the same modulation, by
a differential adjustment of the meter -circuit resistances. These meters
were also used in bringing calibrating and directive modulations to
the same level. A differentially connected modulation -level meter, which
13 I-Sect. 3, D, p. 77.
gave a central zero reading when the two systems were equalized in
the calibrating operation, permitted bringing the two directive modu-
lation depths accurately to equality by adjustment of the east-west
amplitude-controlling Variac to give a zero reading during normal
directive operation.
All monitoring instruments were grouped together, as shown in
Figure 8 and more clearly in Figure 14. The cathode-ray oscillographs,
the only monitoring instruments actually operating at radio frequency,
were built into the transmitter proper. The four meters of Figure 12
Fig. 16- Complete aircraft equipment (without antenna).

and the five of Figure 13, with their associated equipment, were mounted
in the separate housing shown in Figure 14.
d-Indicators
For receiving, a vertical rod antenna, shown in Figure 15 mounted
on the top of a transport airplane, was used. The circular ground plate
at the antenna base was for the purpose of reducing interference from
imperfect bonding of the airplane structure. A crystal -controlled
superheterodyne, made for another purpose but having the flat auto-
matic gain control and high audio fidelity which are characteristic of
modern ultra- high-frequency aircraft receivers, was used. Audio out-
put from a receiver phone -jack actuated the indicators, which also drew
364 RCA REVIEW
tube plate power from the receiving dynamotor. Figure 16 shows the
entire aircraft installation, less antenna. The large unit, which is the
receiver, and the next smaller one, which contains the auxiliary appara-
tus for the indicators, were mounted out of the way. The three small
units -a
control box, a deviation -indicating meter, and a cathode -ray
indicator in its cylindrical magnetic shield- constituted the cockpit in-
stallation. Both indicators were mounted in standard aircraft instru-
ment holes.
The indicator system employed in the ultra-high- frequency work
was rather more complex than the one previously used, in the ways
enumerated below. A phase-shifting control was introduced into the
audio line from receiver to indicators to avoid the need of mechanical
rotation of the azimuth indicator scale for zero setting. To reduce the
receiver power output required for indicator operation, an amplifier
was incorporated into the filter for the sinusoidal signal component.
This filtering amplifier was made of the selective- feedback type,14 to
avoid the weight of high -quality low-frequency reactors, and was
equipped with an automatic gain control, to hold its output very con-
stant and thereby relax the requirements on the receiver automatic gain
control. A local impulse generator was incorporated into the impulse
signal channel to improve the shape and equalize the size of the im-
pulses applied to the indicator, as well as to compensate for the wide
difference between the size of impulses received at sinusoidal modula-
tion maximum (when receiver is north of transmitter) and at minimum
(when south of transmitter) .'5 This generator was a quiescent unsym-
metrical multivibrator, operating through one cycle only when trig-
gered by a received impulse. Finally, to improve the appearance of
the cathode-ray indication by substituting a bright notch for the dark
gap formerly used as the indicating mark on the bright circular
cathode -ray trace, a simple radial- deflecting device for actuation by
the impulses was added.16
As most of the circuits of the indicator, while not novel, are a
little out of the run of ordinary radio circuits, they are shown in some
detail in Figure 17, grouped according to function by the dashed lines.
The filtering amplifier with its automatic gain control and the impulse
shaper are strictly auxiliaries ; they may be regarded simply as special,
selective receiver-output stages. The vibrator -type high-voltage supply
for the cathode -ray tube, having a vacuum -tube rectifier and a trans-
former winding to feed the heaters of the tubes with cathodes at high
voltage, is not shown. Had the receiver not happened to have an over-
14 "A New Type of Selective Circuit and Some Applications" by H. H.
Scott, Proc. I.R.E., Vol. 26, No. 2, p. 226, Feb. 1938.
15 I-Figure 2.
16 I -Sect. 2, B, p. 68.
().11,V'/1 /kh,'('T/UNAL RAU/U -RANGE SYSTEM :3E5
-11
+ p
'13::
Q/ 1
N>3
- n
'00'
1 cor '11
V
1
z
r 0- 0-oo 's I
--
a-1l-4
J
r
I,
J I
3GG RCA REVIEW
size dynamotor, the vibrator supply would have been arranged to pro-
vide plate voltage for the above auxiliary tubes also.
To the right of the dot -dash line in Figure 17 appear the indicators
proper. Placed ahead of the filter amplifier to avoid unproductive load-
ing of the latter, the zero-setting phase control really belongs func-
tionally in this part of the picture. The resistance -capacitance network
between this control and the first amplifier grid served to keep impulse
voltages out of the sinusoidal- current circuits and to compensate nor-
mal phase displacements in the receiver audio circuits and filter, so as
to permit the control to work near the center of its 15-degree range.
Following the filter, a resistance- capacitance bridge was used to give
the 90- degree phase shift necessary to produce a circular trace on the
cathode -ray -tube screen ; the four deflecting plates of the tube were
connected to the four corners of this bridge network with a normal
centering control for each pair of plates, and with sensitivity -equalizing
potentiometers for one pair of plates. To the four corners of the bridge
were also connected four equally spaced fixed taps on an ostensibly
uniform, high -resistance carbon -composition ring. Two diametrically
opposite, freely rotatable brushes bearing on this ring permitted pick-
ing off a voltage, balanced to ground and of reasonably constant ampli-
tude, which could be continuously varied in phase throughout a 360 -
degree range. This control was used to set the deviation indicator to
any desired course.
Operation of the radial deflecting system was directly similar to
that of the familiar Heising constant -current modulator used in trans-
mitters. The high- voltage supply fed the accelerating train of the
cathode -ray tube and the plate-cathode circuit of a pentode, in parallel,
through a common resistor. Normally cut off, the pentode had most of
the time no effect on the operation of the cathode -ray tube. When a
momentary positive voltage from the impulse shaper was applied to
the pentode grid through a high -ratio isolating transformer, the pen-
tode plate current in the common resistor dropped the accelerating
voltage applied to the cathode -ray tube. This slowed the electrons in
the cathode -ray beam and made them momentarily easier to deflect.
The impulse thus appeared on the fluorescent screen of the indicator
tube as an outward- pointing notch or jog on the bright circle of the
normal cathode -ray trace. To compensate the dimming tendency of the
extra radial spot velocity during the jog, a positive impulse voltage
was applied to the control grid of the cathode -ray tube. This voltage
was obtained by by-passing a portion of the common voltage- dropping
resistor to the cathode of that tube. Focus was maintained during the
jog by equalizing the time constants resulting from stray shunt capaci-
tances, and by adjustment of a small trimming capacitor connected
between first and second anodes of the cathode -ray tube. Careful shield-
ing was necessary to prevent the impulse voltages on the accelerating
circuits of the cathode-ray tube from affecting the deflecting circuits.
Not the least of the features of the cathode-ray indicator system
was the straightforwardness of the routine by which it was initially
adjusted to give clear accurate indications at all parts of the azimuth
scale. The indicator trace, before setting -up adjustment, was gener-
ally a defocused, decentered tilted ellipse. After rough preliminary ad-
justments to give proper pattern shape, size, focus, and brilliance were
made, final adjustment for accurate indication was done as follows.
With the range transmitting a calibrating signal with eight impulses
per cycle, the zero- setting phase control was adjusted to give north and
south indications in error by equal amounts and the east -west center-
ing control was then set to give north and south readings of exactly
zero and 180 degrees. Adjustment of the NS phase control to give
equal errors at East and West was then followed by adjustment of the
NS centering control to give east and west readings of exactly 90 and
270 degrees. Finally, the EW amplitude control was set to give the
best average reading at the 45, 135, 225, and 315 -degree points. Since
there was very little interaction between these controls, adjustment
with eight equally timed impulses per rotation cycle was easily made
to bring the eight marks then visible to read exactly correct at the
four cardinal points and correct on the average at the four primary
intercardinal points of the scale. With commercial cathode -ray tubes
the correctly adjusted pattern usually had a slightly elliptical shape,
but because of the method of adjustment used, this did not seriously
impair accuracy of indication at any part of the scale.
Zero-setting and deviation -indicator course- setting controls were
the only ones, except for a power switch, brought out on the face of
the control box (a third knob visible in Figure 16 actuated a manual
volume control, later superseded by the filter -amplifier automatic gain
control) . The other controls affecting accuracy and appearance of in-
dication were available under a dust cover on this unit. It will be seen
from Figure 17 that all elements affecting readings were of a simple,
stable sort and, except for the zero setter, were directly associated with
the cathode -ray tube. The zero setter itself had to be used to compen-
sate any phase -shift variations in the receiver and filter. While it
was the custom to check and trim all adjustments frequently in the
test work, it appeared that all the covered controls were very stable
when not disturbed and that even the zero setting needed little atten-
tion during steady operation. The automatic calibrating transmissions
at quarter-hour intervals gave a more than adequate zero check, and
brief eight -mark periods forming part of these transmissions served to
3(i8 RCA REVIEW
show up and identify any inaccuracy which might develop in any of

the other indicator adjustments.
To actuate the deviation indicator, sinusoidal voltage of freely
variable phase, taken from the tapped -ring course -setting resistor de-
scribed earlier, was applied in push -pull to the suppressor grids of two
pentodes, as may also be seen from Figure 17. Impulsive voltage, taken
from the accelerating train of the cathode -ray tube, was applied to the
control grids of these tubes in parallel. The control grids were biased
somewhat beyond cutoff and swung to zero grid ; excess impulse volt-
age beyond that point was taken up by grid- current drop in a large
series resistor. The suppressor grids were biased to the middle of
their negative control range and were also fed through series resistors
of high value. Their control of plate current was thus limited, on the
one hand by cutoff and on the other by grid- current drop in the series
resistors. This arrangement gave high phase sensitivity near the zero-
point of the sinusoidal voltage, but avoided meter overloading for wide-
ly different phases.
A zero-center microammeter connected between the plates of the
two pentodes, which had separate plate-load resistors, indicated the
differential impulse current through these resistors, while a capacitor
shunting the meter protected it from mechanical damage by the high
peak currents. Because an impulse occurring at either of the two in-
stants of zero sinusoidal voltage during each cycle would give equal
plate currents and zero meter deflection, the sinusoid -phase control
used for course setting was provided with a rough azimuth scale to
avoid ambiguity. Because final course settings were made by bringing
the meter to zero when the cathode -ray indicator showed the craft to
be on the desired course, the accuracy of this scale was unimportant.
Another result of this initial zero setting was that differing normal
plate currents of the two pentodes were not directly harmful. However,
to minimize the disturbing effect of noise impulses, it was found to be
desirable to adjust the two load resistors to compensate for tube differ-
ences. This was easily done by using the eight- impulse -per -cycle trans-
mission, the adjustment then being made for a zero meter deflection
which was independent of sinusoid phase.
Weight is, of course, a major consideration in all aircraft equipment
and the weight of our experimental apparatus was probably greater
than it should be. An ultra-high- frequency receiver will be necessary
equipment in any case; its weight need not be charged to the omni-
directional range. The weight of the deviation indicator with its cable
was 1.8 pounds and is probably irreducible, as is the 3.3 -pound weight
of the cathode -ray indicator with its 9 -wire cable from the control box.
The control box itself weighed 3.8 pounds and, while some reduction
might be possible, it is probable that the total weight of the cockpit

installation of control box, two indicators and their cables cannot be
much less than 8 pounds. The auxiliary unit of the system, with its
multiconductor cable to the control box, weighed 23.2 pounds. Here,
design changes might make an appreciable reduction. In particular, if
the auxiliary equipment were built integrally with a receiver, the added
weight could be made much less than the above value. The degree of
refinement and complexity justifiable in the indicator system can only
be determined by experience in service ; it is difficult to see how it can
be much simpler than the early form described in Section 2A, d.
MEASUREMENT OF THE SLOPE AND DURATION
OF TELEVISION SYNCHRONIZING IMPULSES :::
BY
R. A. MONFORT AND F. J. SOMERS
Engineering Department, National Broadcasting Company, New York
-
Summary Satisfactory operation of television receivers in
requires that the waveform of the transmitted synchronizing signalsthe field
be held
to narrow tolerances. It is therefore essential that suitable measuring
ment and techniques be available at the transmitter so that synchronizingequip-
waveshapes can be accurately and rapidly checked. This paper describes
several measurement methods which have been found to be satisfactory
under ,Tactical operating contlitioas.
1- GENERAL
THE operation of a television broadcast system on a regular public
service schedule over a period of years has shown the need for
various auxiliary measuring and testing equipment. Suitable
equipment and methods for insuring that the synchronizing signals
conform at all times to the prescribed standards are important factors
in maintaining a high- quality service.
The tolerances prescribed for the steepness of wavefronts as well
as
the duration times of the various impulses that go to make up the com-
posite synchronizing signal require measuring equipment capable
of
determining fractional microsecond time intervals. A further require-
ment, equally as important as high precision, is that the measuring
equipment should be simple, and easy to use under actual operating
conditions.
The purpose of this paper is to describe apparatus and methods
developed by the National Broadcasting Company for synchronizing
-
signal measurements. These are the outgrowth of experience
gained
in our television broadcast operation and have been found to
be a satis-
factory solution to the measurement problem.
2- CHARACTER OF MEASUREMENTS
Figure 1, which is a copy of Drawing II of the Federal Communi-
cations Commission Standardsf, will illustrate the character
of the
necessary measurements. These fall into two general classes: Slopes
of leading or trailing edges, and the duration times of the various
impulses. It will be noted that these measurements are to be made
at
the 10 per cent and 90 per cent amplitude points of the wave. This is
to avoid any ambiguity which might arise due to the rounded corners
on the waves resulting from finite bandwidth limitations.
* Presented at Pacific Coast I.R.E. Convention,
Seattle, Aug. 20, 1941.
Docket No. 5806, issued May 3, 1941. "Television Synchronizing
Waveform for Amplitude Modulation."
370
MEASUREMENT OF TELEVISION SYNCHRONIZING IMPULSES 371
An idea of the precision required may be gained from the fact that
under the present standards of 525 lines, 60 fields, interlaced, the maxi-
mum time allowed for the transition from the 10 per cent to the 90
per cent amplitude point of the horizontal synchronizing wave is only
0.25 microsecond. The tolerance allowed for variations in the width or
duration of horizontal synchronizing impulse is also a small value,
being limited to 0.64 microsecond. Similar close tolerances have been
specified for other parts of the wave.
TELEVISION SYNCHRONIZING WAVEFORM FOR AMPLITUDE MODULATION

L LL
w00Rra 1.04 1
ADL<IG 1[ *

a.- A,A A A, RL v.as .w waN
L
J1W1TLIJ11U,rMJ1LLl J, l
I
1 lI l `p) " 11
A u01p4S0
ON- 4[Mrl
i Mwillirl
x
4M 11
.[I
:
1
4 hl l 1i
; 1.
1
I f 11 11 11
II
I 1
RM ..1.1"4-
AIM(.M s 71 DUMB
worm. aAmMs (4 071VE COIN'
Ila' aI11M : t . Oonor O. M. TMM
.
ter is RANK
1I^11
I l '. II ' R A
B
11
u
1(
1N
N
nd
11
1
^
.
1 1
;
1
[Y1.r1
( 1
l0 1 1 1
MO-iOMla 7011N0M4 NOT TO OM( 111 A., sE,. (1)

0E
1) Oo1NwN. O 111N1 (11

O O 1W 17001
NM .17.101 IINNOO[N Of
1711(M.
<MW
H y0 17 r1 4TN01
RM [1P[
K. OLtlI[OIO_
Of *`
;
.ON1
(4);
:
\ E aW LEHR
-- : T
GY
-1
t(NO L'YMI(11
DETAIL BETWEEN CC IN a OMN r1. DETAIL BETWEEN Et IN ,
N[ TOT[
ROM
W (4) l.t INDrOf OW OK TO (TUT Of NUT LIN[.
START
(1
I 4 TINE Fpl
STA. Of OM FOLD TO STAIN Of Not KT POW
Ow or ( LOADING MO TAMING (0411 Of vtNTICAL.S1AN1wM 0010 01 030L([ i1
D
LMMNbNO
1.1w11t
wO/ML 17AaN
NA. MPH 4S TWO
L(0010
TO
O,N
MOO
17.IO174(
NA U1
1D
aO.
MAO
OtwOf
vL1rs(.n) .10
01.14
(+I WC(N
MIT
AL1
[
CppnO111 Of
0<VCN
nctuIO m11T[1T
-ia o_ HAT
() _OMN(N([Nou(,) 001N (I)

ool
4 0.ON1014
I011t10 OM"
1171E0 1TI1 Al USTIONI iNOCtl THAT TOL413(3 0140 .E
700 LwG n1( vANU T /0. 3.01[1371, <TU(
V1
'
t ((3[I4(N IO40N,4(.T,CL I(TNAO[SNALL

FO1 ( IO COwL(T( N OO4
DETAIL BETEN D.0 IN B. [PIaIIING .OLIO 501A 011 IO DE A( 1O O( Of T1O AO" Of
Nd1t0N11 1171b1 11LIO
Fig. 1 -Copy of standards.

Accurate measurements of this kind require the use of an indicat-
ing device having negligible inertia and a wide band frequency re-
sponse. A cathode -ray oscilloscope or its equivalent is therefore gen-
erally employed. It can be shown (see Appendix I) that faithful
reproduction of the wave of Figure 1E which has equal build -up and
decay times of 0.5 per cent of a scanning cycle requires that the
oscilloscope signal amplifier be responsive up to the 200th harmonic of
the scanning -line frequency. Translated into terms of the present tele-
vision standards, this means that the oscilloscope signal amplifier
should have linear phase and flat amplitude response from below 60
cycles up to at least 3.15 megacycles. In addition, it is also required
that the oscilloscope time -base amplifiers be free from compression or
372 RCA REVIEW
non -linearity over the range of sweep amplitudes used in making meas-
urements. Finally, in order to observe the finer details of the wave,
a cathode -ray tube having a sharply focused spot and a large -diameter
screen should be utilized. A tube having a screen diameter of at least
five inches should be employed, but a nine -inch tube is preferable for
accurate work.
3- INADEQUACY OF SAWTOOTH SWEEP
As might be expected, ordinary methods of measurement fail when
applied to the specialized waveforms of Figure 1. A common example
of this is illustrated by Figure 2 which shows the measurement of
ti
PUI sE VYrA, =1 = b x /00/,

ScoPE of-
LEALVN6 EGGE -S = D N ICk)%1
Fig. 2- Sawtooth sweep measurement.

impulse widths and slopes using the usual oscilloscope sawtooth wave
time base. The oscillogram shows one complete cycle and part of
another. This is necessary since the finite retrace time of the sawtooth
wave makes it impossible to see a single complete wave on the forward
trace unless the sawtooth wave is a submultiple frequency of the wave
being measured. In this procedure the dimensions C, C', and D would
ordinarily be read on the oscilloscope screen with a transparent milli-
meter scale and the widths and slopes in percentage values would be
C
Width = W = -X
D
100 per cent
C'
Slope = S = -X 100 per cent
D
On a five -inch oscilloscope screen, the necessity of showing at least
two cycles means that D will be less than 100 mm. At the same time,
the probable error in scale reading, taking account of the width of the
oscilloscope trace may easily amount to 1 mm. It can readily be seen
that serious errors will occur if attempts are made to measure the slope
of the leading edge of the horizontal synchronizing wave with this
setup, since the value C', being less than 0.4 mm, cannot be accurately
read with the scale. As a matter of fact, errors due to unsteadiness of
the image and non -linearity of the sawtooth wave can easily add up to
SAN # Z
L
c
e
% WhOTH-/OV%= 3,x/00,6- /ox/00Z
/v won,/
/8
Fig. 3- Sine-wave measurement of widths.
more than 1 per cent of the scanning cycle, making the error greater
than the value to be measured.
In view of these errors, it has been necessary to develop more
accurate methods, keeping in mind the requirement that the equipment
should be simple, and easy to use under practical operating conditions.
4- SINE -WAVE HORIZONTAL SWEEP

A method of measurement which is at once simple and accurate
within reasonable limits, uses an oscilloscope with sine -wave horizontal
374 RCA REVIEW
deflection. The frequency of the sine wave used is either
the same or
a harmonic frequency of the impulse being measured.
The pattern on
the oscilloscope screen is stationary since the sine wave is
derived
either from the synchronizing generator or directly from
the syn-
chronizing impulse itself. A convenient phase-shifting device
is pro-
vided so that the part of the wave of greatest interest can be
moved to
the center of the screen where it appears expanded horizontally.
This
feature of sinusoidal sweep tends to increase the accuracy obtained
in
measurement of impulse widths and slopes, since the dimensions
to be
measured are larger for a given width of horizontal time axis
than if
sawtooth waves are used. Also, since only one cycle appears
on the
screen, the effective horizontal time base can be wider than
with saw-
tooth sweep, where, as already pointed out, more than one
cycle must
be included in the horizontal direction.
The procedure in making measurements with sine-wave horizontal
deflection is illustrated by Figure 3A, which shows a train of
horizontal
driving impulses as they appear on the oscilloscope using 15,750 -cycle
sine-wave sweep. This type of impulse is utilized for energizing
the
horizontal sawtooth -wave generators of television cameras and
video
monitors in the television plant, and normally has a time duration
of
6 per cent of a scanning cycle. The impulse is moved
to the center of
the screen by means of the phase -shifting device already mentioned
and its width or duration can then be found by means of the follow-
ing equations, derived with the aid of Figure 3B.
Sin - = D2
O
2
C
D
2
length of arc y
Per cent width = X 100 = X 100
circumference 360
B
2
Per cent width = X 100
180
Per cent width =

Sin -' cll
D , where the angle is expressed in degrees.

1.8
By measuring C and D in mm, with a transparent scale attached
to
the oscilloscope screen and applying the above equations, the duration
of the impulse can readily be found. Calculations may be eliminated by

the use of the nomographic chart of Figure 4, which gives the duration
of the impulse directly in per cent of a scanning cycle. Figure 5 is a
photograph of the composite television signal as it appears on the
oscilloscope for a 15,750-cycle sine -wave sweep. The lower -case letters
refer to the impulses as shown in Figure 1. The width of any of the
impulses indicated may be determined by shifting them to the center of
the screen, measuring C and D, and referring to the chart. The width
90 r9
i
r
+
80 8
is
70 7
co c
S
CA
3D S
10
t0
40 4
9
D
ISO 30
_I 3
SINE _DC'
4
T I. 8 (Foa 0J. TO 37;)
.31141 -'
8
(F0237; TO 50,0 b
170 IMPULSE ME4.5UPEMENT CHAQT

FOR MEASUQ /NC /MPOL SE W/OTHS /N FE2CEN7;
US /NC J /NE WAVE HOP /ZONTFJL DEFLECT/ON ON
THE CATHODE Rh'Y 0SC /LLOG2QPH USE
EAPANDED SCA[E
180 0 0
Fig. 4-Nomographic chart.

of vertical blanking may be found as shown in Figure 6, which is a
photograph of the composite signal viewed on 60 -cycle sine -wave sweep.
The same chart may be used for this measurement, the result being
expressed in per cent of one field- frequency scanning cycle.
One of the most serious sources of error in making measurements
of this kind is in taking the millimeter scale readings. It can readily
be shown, however, that errors in scale readings are minimized when
sine -wave sweep is used, as compared to linear or sawtooth sweep. To
give a concrete example, suppose we are measuring the width of the
376 RCA REVIEW
horizontal synchronizing impulse which has a specified duration

of 8
per cent 1 per cent of a scanning cycle and that the D dimension
is
100 mm. For the sawtooth sweep, the value of C will be 8 mm.
However,
the error in reading the scale can easily be as much as 1 mm. Conse-
quently, reading D 1 mm too small and C 1 mm too large, gives
an
answer of 9.1 per cent which represents an error greater than the allow-
able tolerance for the width of the wave. Under the same conditions,
using sine -wave sweep with D also 100 mm, reading D 1 mm too small
and C 1 mm too large, gives an answer of 8.42 per cent which is within
the prescribed tolerance. Another source of error is the presence of
FIG.S.
Fig. 5 -Photo of composite signal on 15,750-cycle sweep.
harmonics in the sine wave. While a rigorous analysis has not been
made for all possible harmonic combinations, it may be stated that a 1
per cent 2nd harmonic content, depending on its phase, can produce a
maximum error of 0.22 per cent of a scanning cycle in the measure-
ment of the width of the horizontal synchronizing signal. This amount
of error will probably not be exceeded if the arithmetic sum of all
harmonics present is held down to 1 per cent.
Another source of error is non -linearity or compression in the hori-
zontal sweep amplifier of the oscilloscope. This should be investigated
for the particular oscilloscope used, and measurements should be con-
fined to the range of amplitudes within which the sweep is linear.
If reasonable care is taken to minimize the errors referred to, the
sine-wave method gives impulse widths to an absolute accuracy of 1/2
per cent of a scanning cycle. While higher accuracy is desirable, the
results are at least practical, since they are within the tolerance pre-
scribed for these waves.
Increased accuracy in the measurement of the slopes of leading and
trailing edges can be obtained by using a sine wave which is a harmonic
of the scanning -line frequency. This stretches out the C dimension so
that more accurate determinations can be made. The appearance of
the synchronizing signal when viewed with 157.5 -kc sine -wave deflec-
F/G. 6
Fig. 6 -Photo of composite signal on 60 -cycle sweep.
tion is shown on Figure 7. Since this is the 10th harmonic, the C
dimension is stretched out by a factor of ten. The same equations still
hold and the same chart can be used, the answer obtained being divided
by 10 to get the slope in per cent of a scanning cycle. Observing the
same precautions as outlined for measurements with fundamental-fre -
quency sine waves, impulse slopes may readily be measured by this
method to an accuracy of -0.05 per cent of a scanning cycle, which is
well within the prescribed tolerances.
5--DIRECT-READING SINE-WAVE METHOD

Increased accuracy can also be obtained with the modified measure-
ment procedure shown in Figure 8, which eliminates the need for
378 RCA REVIEW
charts or scales and gives impulse -width and slope values read directly
from a calibrated dial. In this arrangement, the wave is brought into
approximate position with respect to the cross hairs by means of the
coarse phase adjustment. A fine phase adjustment provided with a dial
calibrated in per cent of a scanning cycle is then used to shift the wave
back and forth with respect to a fixed hair line in the center of the
oscilloscope screen. The difference in dial readings for the 90 per cent
amplitude point of the wave at position "A" and the 10 per cent ampli-
tude point at position "B" gives the slope of the leading edge of the
impulse directly in per cent of a scanning cycle. The slope of the
/ - LEA0//,'6 EOv'E /AugiZ SYNC.- - - 2 or m.4sL /N E06Z

3 w
" OOD b69
EYENEGtiK4 - -
- - - f
-G
ow vF.ar, sEc. - - 9
FE..YJFIPT sec - - /O - 0.
D
Fig. 7- Synchronizing signal viewed on 157.5 -kc sweep.
trailing edge can be measured in the same way. Similarly, the duration
of the impulse can be found by taking two readings at the intersection
of the 10 per cent amplitude line and the vertical index mark. The
phase- shifting arrangement shown is a modification of that described
by Hartshorn' and gives a constant output voltage within the limits of
voltage regulation of the amplifier feeding it. Over the small range
required of the calibrated phase shifter, R and C can be chosen so that
there is negligible change of amplitude with change of phase. Tests
show that the results obtained with this arrangement are independent
of moderate amounts of compression in the horizontal deflection ampli-
fier of the oscilloscope, since even under these conditions the sweep will
be linear in the central portion of the screen where observations are
MEASUREMENT OF TELEVISION SYNCHRONIZING IMPULSES 3'79
made. This latter feature plus the elimination of scale-reading errors

makes for increased accuracy.
6-APPLICATION TO TIME -DELAY MEASUREMENTS

An interesting application of the arrangement of Figure 8 is its
use in determining the time delay of networks or lines carrying syn-
chronizing impulses. The wave at the input to the network is lined up
with the cross hairs on the oscilloscope, the setting of the fine phase
CROSS HAIRS
OR MARKS AV
SCREEN
%ANPL
PosiTioiv 4
POS /T/ON - I
/07 PL
/NP//T
O
SAE WAVE-
GEN
F/NE0Add CAL. (SYNC INPUT
/N Y OFSCANN/NC
COARSE O CYCLE ---If
ADJ
Fig. 8- Direct reading sine -wave method.
adjustment being noted. The oscilloscope vertical plates are then con-
nected to the output of the network and the phase shifted to bring the
wave back to its original position. Knowing the frequency of the sine
wave and the phase shift in per cent of a cycle, the time delay in
microseconds can easily be found. This same kind of delay measure-
ment can, of course, also be carried out by the arrangement previously
outlined for impulse width measurements using a millimeter scale and
the chart of Figure 4.
380 RCA REVIEW
FIG. 9A FIG 98
Fig. 9-Interlace check.
7 -USE OF SINE WAVE IN CHECKING ACCURACY OF INTERLACE
Line -frequency sine-wave sweep has also been found useful as a

means of testing accuracy of interlace. For example, if 60 -cycle driv-
ing impulses from the synchronizing generator are applied to the
vertical plates of the oscilloscope and 15,750 -cycle sine wave is applied
to the horizontal plates, the pattern shown in Figure 9A will be
obtained. It should be possible by shifting phase of the sine wave to
cause these pairs of lines to cross each other in the center of the hori-
zontal trace as shown in Figure 9B. If the lines cross anywhere else
except the center, pairing of the scanning lines on the television picture
(lack of interlace) will be indicated.
8- 15,750 -CYCLE SINE -WAVE UNIT

The circuit diagram of the unit for generating the 15,750-cycle
H.D. OR SYNC.
-0 /NRUT
&AB 7 6AC7
-b
PHASE SH/FT UN/T
OV>-
Fig. 10- Circuit of 15,750 -cycle sine -wave unit.

MEASUREMENT OF"TELEVISION SYNCHRONIZING IMPULSES :381
o----
is7soN a
COARSE 9(ADJ p'
6AC7
L_
Cz
6AC7
L9 4
_ G+
NPUT r
f 0+
Fig. 11- Circuit of 157.5 -kc sine -wave unit.
sine wave is shown in Figure 10. This consists essentially of an ampli-
fier tuned to the scanning line frequency, with a low pass filter in its
output. The filter is made up of one constant -K low pass section, an
M- derived intermediate section designed for an infinite attenuation
point at the second harmonic (31,500 cycles) and a terminal half-
/57,37(C PHASE-
A. .FC. /S7,S/KC
OSG, SHIFT OUTPUT
3/, Soo", S.'/

A.F.C. FREQ
OSC, D/V/DER
SYNC GEN.
FREQUENCY
D/V/DER
3/,SOOru
TO
60",
PULSE
O/SCR/M
/NA TOR 60/w 4/NS
Fig. 12- Alternative meLhod of obtaining 157.5 -kc wave.

382 RCA REVIEW
section designed for M = 0.6. The filter impedance is 1000 ohms

and
the cut -off frequency is 17,000 cycles. One type of variable phase
-
shifting network is also shown.
When measuring locally generated synchronizing pulses, this unit
is actuated directly from the synchronizing signal generator.
For sig-
nals originating at a remote point, as is the case with our
mobile
pick -up units, the sine -wave generator is energized from
the syn-
chronizing separator circuit of a local television receiver picking up
the
radiated signal. Measurement information can thus be made available
GAG7 . /552 6AG7
o0
Fig. 13 -Dot frequency generator.
to the mobile units via the order wire or radio link provided
for com-
munication purposes. This is a decided advantage, since space
limi-
tations make it impractical for the mobile units to carry a great
deal
of auxiliary measuring equipment.
9- 157.5 -Kc SINE -WAVE UNIT

One method of generating the 157.5 -kc sine wave is
shown in Fig-
ure 11. This consists essentially of a harmonic generator
and amplifier
tuned to 157.5 kc and driven by the line- frequency
sine -wave unit
already described. To avoid serious amplitude modulation
in the
L,C1, L_,C. combination, coils having a Q in the
neighborhood of 250
would be required. At this frequency such coils would be rather bulky.

A practical answer was found in the use of small spool -wound coils, of
30 No. 36 litz, having a Q of 90. A signal of about 50 volts across L2C2
is fed to the grid of 6AC7/1852 which is operated with low screen
voltage and zero bias to give limiter action. The plate of this tube
feeding a second pair of tuned circuits delivers a signal having only a
negligible amount of amplitude modulation. A slight frequency modu-
lation results from the above process, but checks indicate it does not
appreciably affect the accuracy of the measurements.
Fig. 14 -H. D. modulated by dots.
An alternative method of obtaining the 157.5 -kc sine wave is shown

in block diagram form in Figure 12. This arrangement requires
switching off the regular 31,500 -cycle afc controlled oscillator of the
synchronizing generator and locking in the 157.5 -kc oscillator with
the 60 -cycle mains as shown. This method is suitable for test purposes
only, as a switch of this kind could not be made during a program with-
out causing a disturbance.
10-DOT MEASUREMENTS
An alternative method of making the width and slope measurements
already outlined is to modulate the grid of the oscilloscope with a high
harmonic of the horizontal scanning frequency. Figure 13 shows the
:384 RCA REVIEW
circuit of a harmonic- generator unit for producing the 100th harmonic

of 15,750 cycles. When this signal is applied to the oscilloscope grid,
the wave is broken up into a series of dots as shown in Figure 14. In
this case, one dot plus one space equals 1 per cent of a scanning cycle.
Percentages can be read off directly, no calculations being necessary.
An outstanding advantage of this system is that it gives
results
VIDEO //PUT
V/0E0 nvIONiTO/)
VIDEO AMP
AND
K/NESCORE
SYNC /-/OR/Z VERT

SER OS C. OSC.
/4750-1, S/NE ITAVE

/NFr/r 17
r,
sHiFr
60^ S/NE WAVE
IIIf
NPUT
9iSH/FT
Fig. 15-Block diagram of pulse cross unit.
independent of sweep waveform or linearity. The wave shown in Fig-

ure 14 is a horizontal driving impulse having a 1/2 per cent slope on its
leading and trailing edges. It is obvious that a larger number of dots,
say 500 per scanning cycle, is needed for accurate measurement
of
steepness of wavefront of waves of this sort. At the present time,
how-
ever, production of higher numbers of dots of sufficient stability
re-
quires removing the regular afc controlled 31,500 -cycle oscillator from
the synchronizing generator and switching in a stable fixed-frequency
oscillator in place of it. If this is done, the 500th or higher harmonic
of the line frequency will be sufficiently steady with respect to
the
fundamental to allow very accurate width and slope determinations

to be made.
11- PULSE -CROSS UNIT
While the equipment already described affords means of accurate
measurement of various features of the synchronizing signal, it does
not provide a check on such other items as the number of vertical
Fig. 16A -Pulse -cross correct.
synchronizing sections, number of equalizing pulses, etc. This infor-

mation can readily be obtained by means of the arrangement shown
in block diagram form in Figure 15. In this setup, which is similar
to one described by Loughren and Bailey2, sine waves of 60 cycles and
15,750 cycles are used to lock in the blocking oscillators provided for
horizontal and vertical deflection of the video monitor. Phase shifters
are arranged so that the horizontal and vertical blanking signals can
be made to form a "pulse- cross" in the center of the kinescope screen.
By applying the synchonizing signal to the kinescope grid in the proper
polarity and increasing the vertical deflection, the pattern of Figure
16A is obtained. The values identified by the lower -case letters refer
to the impulses shown on Figure 1. Several incorrect patterns resulting
from faulty adjustment of the synchronizing generator are shown in
Figures 16B, 16C, and 16D.
1
386 RCA REVIEIV
Fig. 16B -Pulse -cross incorrect.
While this pattern gives information as to the number and relative

timing of the various impulses, it is not suitable for the accurate
Fig. 16C -Pulse -cross incorrect.

determination of impulse widths and slopes since the 10 per cent and
90 per cent amplitude points specified in making these measurements
cannot be found accurately on the pattern.
12- CONCLUSION
In conclusion it should be pointed out that, while the measurement
methods presented here have been explained with respect to the specific
synchronizing waveform of Figure 1, they are sufficiently generalized
Fig. 16D -Pulse cross incorrect.
to be useful for other types of television synchronizing signals which

have been proposed.
13- ACKNOWLEDGMENT
The authors wish to acknowledge the help of other members of the
NBC Television Engineering Staff whose constructive suggestions and
active cooperation contributed much to the success of this project.
APPENDIX I
It has been shown by Von Ardenne3 that the general equation of a
sawtooth wave having a finite retrace time Ta which is P per cent of
the period T of the wave is
n =
f(t) =EA Sin nWt
71 -1
388 RCA REVIEW
Ak
where A- ( -1)i+ 1 E- --
)127'
X
100
P
X Sin
(nPir)
(100)
(1)
and Ak is the amplitude of the sawtooth wave.
The amplitudes of the harmonics decrease more rapidly than 1/n and
periodically become zero, the first zero position lying at the value of n
which equals 100 P. Inclusion of harmonics lying beyond the first zero
contribute little to the shape of the sawtooth wave except to reduce the
slight rounding at the peak of the wave. Therefore, the sawtooth wave
.G".n.' .M-
ki 30
poi
t'
A;
;ta- E
..._.
. fBurv-aPit 7",cs 7/7,1 //1/S/6
..
Fig. 17- Diagram for Appendix I.
is faithfully represented to a very close approximation by the summa-

tion of harmonics up to and including the first zero.
By a simple extension of this theory, it may be shown that the
same conditions hold true for the number of harmonics necessary to
represent a square- topped impulse type wave for which the build -up
and decay time Ta are both equal to P per cent of the period T. In
Figure 17 is represented the plot of two sawtooth waves of equal
periods and times of decay, the wave B being delayed a time S seconds
behind the wave A. As shown in the diagram, subtraction of wave B
from wave A yields an impulse -type wave having a period T, the

build -up and decay times, t, both being equal to P per cent of the
period. The general equation of the resultant wave may be expressed
as given in Equation (2), the value of A being identical with that
above.
n- o0
f (t) =)-- A [Sin nwt - Sin no,t (1 + 6)] (2)

n -1
The new wave has the same period as the sawtooth waves, its build -up
and decay times Ta are both equal to the decay time of the original, and
it also contains harmonics of identical frequency, and comparable
amplitudes. It, therefore, follows that the impulse -type wave will also
be represented to a close approximation by the inclusion of harmonic
components up to and including the value of n equal to 100 /P.
REFERENCES
1 L. Hartshorn, Proceedings of the Physical Society (London) , Vol. 49,
Part 2, March 1937.
2 "Special Oscilloscope Tests for Television Waveforms," by A. V.
Loughren and W. F. Bailey, presented at I.R.E. Rochester Fall Meeting,
November 13, 1940.
3 "Frequency Spectrum of Sawtooth Waves," M. Von Ardenne, Tele-
vision and Short Wave World, January 1938.
OUR CONTRIBUTORS
ALLEN A. BARCO, a native of St. Louis, Missouri, re-

ceived his degree of B.S. in Electrical Engineering from
Washington University in 1937. Following his gradua-
tion he joined the engineering staff of the Radio Corpo-
ration of America License Laboratory, where he con-
tinues to be engaged. Mr. Barco is an associate member
of Sigma Xi.
EDWARD W. HEROLD received a B.Sc. degree at the Uni-

versity of Virginia in 1930. He was employed in the
research section of the Bell Telephone Laboratories from
1924 to 1927. During 1927 and the summers of 1928 and
1929, he worked in the engineering department of E. T.
Cunningham, Inc. Since 1930, Mr. Herold has been in
the Research and Engineering Department of the RCA
Manufacturing Company at Harrison, N. J. He is a mem-
ber of the Institute of Radio Engineers.
CHARLES N. KIMBALL received a B.E.E. degree from

Northeastern University in 1931. From Harvard Uni-
versity he received his M.S. degree in 1932 and his
Sc. D. degree in 1934. He spent two years with the Na-
tional Union Radio Corporation and since 1936 has been
connected with the RCA License Laboratory. Dr. Kim-
ball is an associate member of the Institute of Radio
Engineers.
DAVID G. C. LUCK was born July 1906 in Whittier,

California. Bachelor of Science Mass. Inst. of Tech. 1927;
Ph.D. 1932. Swope Fellow in Physics 1927 -28, Malcolm
Cotton Brown Fellow 1928 -29. Assistant, Department of
Physics, M.I.T., 1929 -32. Research Division, RCA- Victor
Co. 1932 -35, Victor Division of RCA Manufacturing Co.
1935. Member, American Physical Society.
RAYMOND A. MONFORT attended the University of

Kansas, and later, New York University. From broad-
cast station operator in Kansas City, he went to work for
Western Electric in Kearny, New Jersey as Equipment
Engineer. In 1932 he joined the National Broadcasting
Company. Since 1935 he has been identified with NBC
Television, and is now Television Maintenance Super-
visor in the New York Studios. He is an associate
member of the I.R.E.
390
OUR CONTRIBUTORS 391
FREDERICK H. NICOLL received a B.Sc. degree in Physics

from Saskatchewan University, Canada, in 1929 and an
M.Sc. in 1931. He held an 1851 Exhibition Scholarship
to Cambridge University for three years research and
received a Ph.D. from that university in 1934. He was
a research physicist with Electric and Musical Industries,
Ltd., in London from 1934 to 1939. Since that time he
has been with RCA Victor Division of the RCA Manu-
facturing Company, Inc., as a research engineer. Dr.
Nicoll is an Associate Member of the Institute of Radio
Engineers.
GEORGE M. NIXON attended Pratt Institute and New

York University. Since 1928 he has been engaged in
general development work in the Development Group of
the National Broadcasting Company. Mr. Nixon is a
member of the Acoustical Society of America and the
American Institute of Electrical Engineers.
DWIGHT O. NORTH received his B.S. degree from Wes-

leyan University in 1930 and his Ph.D. degree from the
California Institute of Technology in 1933. Since 1934,
Dr. North has been with the Research and Engineering
Department of the RCA Manufacturing Company at
Harrison, N. J., engaged principally in research studies
of tube and circuit noise. He is a member of The Insti-
tute of Radio Engineers and a member of the American
Physical Society.
STUART W. SEELEY received his B.Sc. Degree in Elec-

trical Engineering from Michigan State College in 1925.
He was an amateur experimenter and commercial radio
operator from 1915 to 1924. Following this he joined the
experimental research department of the General Elec-
tric Company, and a year later became Chief Radio
Engineer for the Sparks Withington Company. Since
1935 he has been Section Engineer in the RCA License
Laboratory.
FRANK J. SOMERS was born at San Jose, California,

May 27, 1908. From 1922 to 1928 he was a licensed
amateur radio operator. During the summer vacations
of 1928 and 1929 he worked as a laboratory assistant
;a. ic... on long wave receiver design for the Federal Telegraph
Co. Receiving the degree of B.S. in E.E. from the
University of Santa Clara in 1930, he joined the Bell
Telephone Laboratories as a member of the technical
staff in the Submarine Cable and Special Research
Department. From October 1932 to June 1933 Mr.
Somers attended the Graduate School of Electrical Engi-
neering at Stanford University. From September 1933 to March 1935 he
was employed as a research and development engineer by Television Labor-
atories Ltd. at San Francisco, Cal. From March 1935 until 1938 he was
engaged in design and development work on U -H -F. Transmitters, tele-
vision receivers and studio equipment for Farnsworth Television, Inc., at
Philadelphia, Pa. From 1938 to date he has been a member of the Tele-
vision Engineering Staff of the National Broadcasting Co. Mr. Somers
is an Associate Member of the I.R.E., and is a licensed Professional Engi-
neer in the State of New York.
TECHNICAL ARTICLES BY RCA ENGINEERS
Published Fourth Quarter, 1941
CAR?o N. It. It. .\ "I I,o-nle \.m- Turnover Automatic Record ('hanger -RCA Review,
.
October.
DEAL, HAN Ai B. Recei%Tr Control by Transmitted Signal
( tctober. "Alert Receiver " -RCA Re;.,,
-
1)1 1.1-, V No. J. A ,\ leth0I and Equipnwnt for Checking Television Scanning Linearity- -RCA
/icVicw, t)ctuhcr.
I'F:HHI N. W. R. "el.' WAGNER :HUI FERRIS.
FL ET, HER. I.. E. and ('. L. KENNEDY -- A Modern Control Room

mitter Central for a Commercial Radio Trans-
R(C.'I Reri,-a, October.
Fui.KEk IS, IL F. and P. A. RICA ARDS -Photography
October. of Cathode-Ray-Tube Traces -RCA Review,
Fos EH, D. E. and J. A. RANKIN -Intermediate- Frequency

Values for Frequency-Modulated-
Wave Receivers Proceedings of the I. R. E., October.
HILLIER, J. -A Discussion of the Fundamental Limit of Performance of
- Physical Review, October. an Electron Microscope
- see ZwoRYK IN, HILLIER. and VANCE.
Ki- s.sEDY, C. L. See FLETI II ER and KENNEDY.
I:,.( UAN, N. L- -Coupled Resonant Circuits for Transmitters-
October. Proceedings of the I. R. E.,
LI.F.nF:I.I.YN, F. B.- -I3ook Review of "Electron Inertia Effects"-Proceedings of the
October. I. R. E.,
MAI.TF:R, L. -The Behavior of Electrostatic Electron Multipliers as a Function
Proceedings of the I. R. E., November. of Frequency--
.Iolt)oN, G. A. -A Survey of Research Accomplishments \Vitt the
RCA Rct'icuw, October. RCA Electron Microscope -
RANKIN, J. A.- -See FOSTER and RANKIN.
RICHARDS, P. A.- -see FoLKERTS and RICIARDS.
- see WAILER and RICHARDS.
THOMAS, H. E. -The Development of a Frequency -Modulated Police Receiver for Ultra-High-
Frequency l'se- -RC:l Review, October.
'1'110\ esos, B. J. -- Voltage Controlled Electron
November.
Multipliers -Proceedings of the I. R. E.,
\'ANCE, A. W.- -see LwORYKIN, HILLIER, and VANCE.
WAGNER, 11. M. and \V. R. FERRIS-The Orbital -Beam Secondary -Electron
high- Frequency Amplification-Proceedings Multiplier for Ultra-
of the I. R. E., November.
WAILER, L. C. and P. A. RICHARDS-A Simplified
and Experimenter -RCA Review, Television System for the Radio Amateur
October.
ZWORYFIN, V. K., J. IIILLIER, and A. W.
VANCE
of a 300 -Kilovolt Magnetic Electron -A
Preliminary Report on the Development
October. Microscope -Journal of Applied Physics,
392

The Absolute Sensitivity of Radio Receivers - North - RCA-Review-1942-Jan

Caricato da

Informazioni sul documento

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

The Absolute Sensitivity of Radio Receivers - North - RCA-Review-1942-Jan

Caricato da

Copyright:

Formati disponibili

RCA REVIEW

A Quarterly Journal of Radio Progress

VOLUME VI January, 1942 NUMBER 3

Generation and Detection of Frequency -Modulated Waves 269

A New Chemical Method of Reducing the Reflectance of Glass 287

An Omnidirectional Radio -Range System,

Measurement of the Slope and Duration of Television Synchronizing

Copyright, 1942, by RCA Institutes, Inc.

1t.trH It. BtL CHKt r.r, W. IfuKN

vl s. .st, A wt(! Urt'tluputt str, .iaYlaf/lllt l;ruc ru1 l'ulutltl

Enyist, rr tn Chu/ Vt., Chief Engineer,

O. B. HNwN CHKl2, H. Tai

Previously unpublished papers appearing in this book may be reprinted,

TWO new auditorium -type broadcast studios were placed in oper-

Each studio accommodates an audience of about 450 in comfortably

The upholstered chairs, in addition to being comfortable, perform

the tonal quality of the performer or performing group is enhanced.

these surfaces to alter acoustical conditions in the studio, as desired.

likewise untreated except near the stage to permit microphone usage

rear wall of these studios; the small amount of energy reflected is

All necessary openings in the boundary surfaces have been given

ing forms an interesting pattern on the ceiling and provides indirect

through it some of the controls are visible. The general auditorium

concentration on the stage because of the contrast in illumination. The

program as heard directly in th studio from the seating area would

The planning of these studios was under the direction of Mr. O. B.

Summary -This paper contains a description of a frequency detector

PART I-LINEAR FREQUENCY -MODULATED MONITOR DETECTOR

OPERATION of the detector unit may best be described with the

increase the average value of current flowing through Ro. Conversely,

The circuit of the laboratory model of the monitor detector, shown

introduce undesired components into the detector output. Therefore,

approach zero asymptotically as the frequency was increased, and the

EXPERIMENTAL DETERMINATION OF DETECTOR LINEARITY

determined with a precision of about 0.02 per cent (approximate read-

PART II -LOW DISTORTION FREQUENCY -MODULATION GENERATOR

mined principally by curvature in the reactance tube characteristic,

PRINCIPLE OF OPERATION OF EXPERIMENTAL

circuit operated at 1 /x/2 times its resonant frequency. An investiga-

quency of an unmodulated low- frequency carrier (generally about 100

a TO MUL T/PL /CR

equal in magnitude to the reactance of the circuit capacitance, for all

at the desired carrier frequency. The phase deviation is, of course,

linear effects at low audio frequencies (which require the greatest

unlimited multiplication of the phase deviation without resorting to

parallel chain of multipliers, an unmodulated 100 -kc signal (from the

Figure 8 is a diagram of the phase -modulator circuit employed in

Fig. 8 -Phase modulator circuit.

At higher audio frequencies (greater than 1000 cycles) the dis-

MODl4 4TOR / OR/O AMS

audio cycle. However, the rms distortion at 50 cycles is relatively

E,, curve. This can be corrected, if desired, by some types of regen-

SO /00 ;GOO ,0000

AUD/O FQS QuENCY - CYCLES PER SECOND

.4: - s,,, cry,

The circuit is resonant at V2 X the source frequency, i.e., C in the

and (FtC0), 0).2 1 1

2w,2 / je,8C 2RCw, -1- j1 1

The absolute magnitude of 73 is therefore 'z: 1

(2) The phase vs. R equation of the circuit is readily determined

The ratio of coefficients of the 2nd harmonic and fundamental audio

The frequency deviation is equal to

Summary-A new chemical method of reducing the reflectance of glass

THE use of an externally applied transparent film on glass, for